Strategy library
614
The strategies the desk screened
What each rule does, read from its code, and how it did across 13 coins and three timeframes.
Open the strategiesIndicator library
31
Indicators
What each calculation measures, how it is read, where it fails, and how many screened strategies use it.
Open the indicatorsThe study · the dataset, CC BY 4.0
We ran over 600 TradingView strategies on crypto. Here is what survived.
47,242 backtests, a breadth filter, an after-fee replay with credibility gates, and the table of every result, downloadable.
Method4 min
What drawdown is, and why it is the number that decides whether you keep going
Drawdown is the fall from a peak to the trough that follows. It tells you what a strategy would have asked you to sit through, which matters more than the return it would have paid.
Read
Reading claims3 min
What a labeled record looks like, with the desk's own as the example
Three kinds of numbers exist in trading research, and almost every record you will ever see blends them. Here is what it looks like when they are kept apart, figure by figure.
Read
Method3 min
The six-sleeve portfolio: how the desk's method is built, and why it stands aside
CopyOrders' method runs six independent sleeves on equal capital, each with its own regime filter, so that no single idea decides the record and standing aside is a position, not a failure.
Read
Method4 min
How to read a backtest (and when to walk away)
A backtest shows how a set of rules would have behaved in the past. Seven checks separate one you can learn from, from one that was built to be shown to you.
Read
Lesson4 min
Why advertised day-trading results fall apart
A one-off gain is not a growth rate you can compound. A win rate that looked strong in an old sample can flip once the market regime changes.
Read
Lesson4 min
Why cherry-picked backtests fail
A backtest that starts from a chart you already like mostly confirms a bias. A useful test changes the rules and checks whether the result still holds.
Read
Prop firms6 min
What makes a trader fundable
A fundable trader is one a firm can leave capital with overnight rules still intact. A strategy that makes money on a personal account can still fail that test.
Read
Lesson4 min
What rough volatility means for option pricing
Rough volatility means log-volatility moves scale more slowly than Brownian motion. Pricing models that skip that roughness misstate exotic paths.
Read
Lesson4 min
Why two option pricing arguments agree
Black-Scholes hedging and the no-arbitrage expectation look like different prices. Under the model's own assumptions, a physics identity shows they are the same function.
Read
Lesson4 min
When trading is not a fixed-edge game
A dice market quoted at 2 and 4 expects a gain of 0.5 a trade, and paths approach it. Grocery prices and stock trades do not. There the policy has to change.
Read
Lesson4 min
What a network learns from a dice roll
A network trained on dice rolls learns the mean, about 3.5, not the next face. The same object is a conditional expectation, stable for a face and unstable for a stock.
Read
Lesson4 min
How many dice maximize one six
The chance of exactly one six is the same at five dice and at six. A continuous maximum near 5.48 sits between them, so both counts are optimal.
Read
Prop firms4 min
Score the economy, then time the forex zone
Tobias scores growth, the central bank, and risk mood before a zone. His NZD/CAD example used at least a 29-pip stop and a fixed two-to-one target.
Read
Lesson4 min
Why martingale betting blows up in volatility
Doubling after each loss recovers the opening stake only with unlimited capital and no bet cap. A finite bankroll, or a short volatility bet, runs out of room first.
Read
Lesson3 min
Why market timing breaks on real data
A known die lets you trade around 3.5 and compound. After it becomes ten-sided, the old quote loses, and a five can come from either die. The regime stays hidden.
Read
Lesson4 min
How a paper trading bot with an LLM is built
The bot stores symbols, drawdown levels, and an on-off flag, then sends one-share paper orders. A language model reads the book and answers. It does not choose entries.
Read
Lesson4 min
Show live broker data on a Flask dashboard
A Flask page can show a live broker cache by reading the trading object's lists on a POST route. Ajax refreshes that HTML about once a second while the feed runs.
Read
Method15 min
VWAP trading strategy explained
A VWAP strategy uses the volume-weighted average as the fair price for the session. This version stacks three of them and keeps the stop to about ten ticks.
Read
Lesson4 min
Read progress with a rolling mean, not one score
One score is a noisy draw around a moving level. A seven-period mean tracked the practice tests as they rose, while the full-history mean lagged and overstated downside.
Read
Lesson4 min
Code geometric Brownian motion sample paths
A Python loop samples geometric Brownian motion by multiplying the last price by an exponential step. A demo starts at 100, with 30 percent volatility and 100 paths.
Read
Lesson4 min
Which quant models earn their keep
Markov chains top this working list. GARCH and principal components sit just below. Autoencoders and Nelson-Siegel rank lower because they are easy to misuse.
Read
Lesson4 min
Reduce Monte Carlo error with a control variate
A control variate subtracts a known-mean extra from each payoff. The target price stays put, and the lecture's toy run cut variance by about three quarters.
Read
Prop firms6 min
Which strategy passes a prop firm challenge
On one firm's challenge book, pass rates differed by style: reversal, breakout, trend, price action, and order flow. The ranking is that book, not a proof that one style is best.
Read
Lesson4 min
Why expected value is not enough to trade
An edge is a positive expected value, and it fails if one losing streak ends the account. Quants stumble when a backtest replaces live updating.
Read
Lesson4 min
When a backtest ratio is not enough
Win rate, monthly cash, and a high Sharpe ratio describe the past. The lecture's test is whether a live book still resembles that past before crowding or a regime shift.
Read
Lesson4 min
Find the cutoff a normal probability asks for
A critical value is the cutoff with a chosen left-tail probability. On the standard normal curve, the cutoff for 95 percent to the left is about 1.65.
Read
Lesson4 min
Why priced-in does not mean correct
A fast price reaction is not proof the price is right. Alpha is the leftover orthogonal to a risk model, and a home run still required you to be positioned.
Read
Copy trading3 min
Does copy trading work? An honest answer from a desk that is building it
Copy trading works mechanically and fails financially more often than it should, for three reasons that have nothing to do with the signals. Here is how to judge whether it is for you.
Read
Lesson4 min
A capped fund beside free quant education
This plan is a fund under general solicitation, capped at 100 investors, plus free lessons and career material. Partner interviews were only starting.
Read
Lesson4 min
How a utility function frames growth and risk
A policy maps the situation to an action. Utility scores the result, and principles draw that score. Reaching a higher peak means a temporary drop first.
Read
Prop firms7 min
How to pass a prop firm challenge
Four funded traders describe four different ways through a challenge. What they share is a written risk limit, not a shared entry signal.
Read
Lesson2 min
What an empty intro recording contains
The April 6, 2022 recording is only a greeting. It has no lesson, no formulas, and no market result to report.
Read
Lesson4 min
Karhunen-Loeve series for the Brownian bridge
The Brownian bridge is a random path pinned at both ends. Its Karhunen-Loeve series builds that path from sine waves and independent normal draws.
Read
Lesson4 min
Three properties that define Brownian motion
Brownian motion starts at zero, is almost surely continuous, and has independent normal increments. A five-day sketch shows the cloud, and the integral is the end value.
Read
Lesson4 min
Estimate a Poisson rate with maximum likelihood
For Poisson counts, maximum likelihood picks the sample average. A simulated hospital with true rate 85 recovered about 84.8, and a full-bed chance near 5 percent.
Read
Lesson4 min
How a tariff shock shows up in a cash-flow model
A no-tariff cash-flow sketch values the firm at 400. Tariffs cut near-term flows, then the second sketch lets the firm reoptimize. The gap is a model, not a buy signal.
Read
Method15 min
When selling options premium has no edge
Selling a put collects premium that looks like a high win rate. This account says a 20 or 30 delta short put still blows up, and the edge is the risk frame, not the win rate.
Read
Lesson8 min
How Nasdaq effort stacks one-to-one scalps
Fabio Valentini says that Nasdaq effort reads who is dominating, and that several one-to-one scalps fit a drift better than one large target. The win-rate remarks are his, not an audit.
Read
Lesson8 min
How effort versus result filters a gold scalp
Darius Kehlmann says that context, location, and confirmation share one weight on gold. A volume profile is effort and a TPO is time. The payout figures are the host's introduction, not an audit.
Read
Lesson4 min
What a ChatGPT order bot actually does
The bot sends broker news and a price to a model, which returns a thesis, a target, and order arguments. One pass produced a 250 target, then a routing error stopped it.
Read
Lesson4 min
Simulate two correlated Brownian paths in Python
Give two Brownian paths a chosen correlation with a matrix factor or with one formula. A finite sample landed near the target, not exactly on it.
Read
Method10 min
How to trade the range after a strong move
After a strong impulse, price often builds a range. One model fades the edges of that range and stands aside while the impulse is still running.
Read
Lesson4 min
How a bid-ask spread captures the middle
A market maker quotes a bid and an ask around a forecast. If the result lands between them, both bettors lose and the quote captures the spread.
Read
Lesson4 min
What quant trading changes in practice
Everyone sees the same past data. A quant model turns it into an expectation, such as implied volatility rich to realized, and has to be rebuilt when that gap breaks.
Read
Lesson3 min
Fit a line in Python two different ways
Least squares has one best line. A matrix formula and a library fit, run on the same ten noisy points, agreed once both answers were rounded to five decimals.
Read
Lesson4 min
Emoji sentiment and cross-sectional returns
Emoji sentiment, after standard controls, still lines up with later returns in one campus study. One standard deviation maps to more than 5 percent a year there.
Read
Method6 min
MNQ scalping strategy on the five minute
On micro Nasdaq, the long and the short in this record are different systems. The long needs a wick through a volatility band. The short is one candle and a clock.
Read
Lesson4 min
What a mean measures, and what variance adds
A mean is a probability-weighted sum, not the most common outcome. Shifts move the mean and leave variance alone. Correlation rescales covariance onto minus one to one.
Read
Lesson4 min
Advanced projects for quant judgment
Build a trading system you can override, a library for rough volatility and signatures, and a lab that times Python tools against a known option price.
Read
Lesson4 min
Investing from a toy to market risk
Investing is paying for a risky shot at a reward. A fair price is the crowd's guess, and stocks that look unrelated can still share one market.
Read
Lesson4 min
A working reading list for aspiring quants
Treat quant books as a lab: pricing, machine learning, text, and linear algebra, plus three Stoic anchors. Read a chapter, build something, then reopen the book.
Read
Lesson3 min
Updating a volatility model as quotes arrive
A badly tuned mean-reversion model can call a VIX above 60 a 65-billion-year event. A Kalman filter blends that model with new quotes. Noise sets how fast it moves.
Read
Prop firms4 min
What prop-firm access costs before any payout
The firm's promo says three tries can cost about $1,700 before trading gain. It wants end-of-day risk and paid winners. The 5 percent growth path is only a scenario.
Read
Lesson4 min
Implied volatility versus what the path realizes
A 30-day realized volatility roughly tracks a known 20 percent level and lags if the level moves. Implied volatility is the input that matches an option's market price.
Read
Lesson4 min
Five papers behind modern option pricing
Five results shape this map of quant finance: Bachelier, Sharpe on beta and alpha, Black-Scholes, Dupire's local volatility, and Carr-Madan pricing by Fourier transform.
Read
Lesson4 min
Why sample means look normal in markets
Sample averages tend toward a bell curve even when each return does not. A shifting market can still make that fixed curve understate the chance of a losing block.
Read
Lesson4 min
Markov chains for a loan book
A loan cannot become 90 days late in a single month. A Markov chain allows only the steps a month can produce, and a matrix power gives the odds many months out.
Read
Prop firms5 min
How to choose a prop trading firm
Before you pay a prop firm, ask how a losing day is defined, what the challenge fee buys if you fail, and whether a trader who follows the rules actually gets paid.
Read
Lesson3 min
Profitable history versus a tradable edge
A strategy can show a positive average in 2020-2023 and a negative one afterward if the gain distribution shifts. Stable regimes are what this lecture calls tradable.
Read
Lesson4 min
Why a researcher left the quant desk
Isolation, not the math, is why this researcher left a Bloomberg desk at 22, then a PhD and an unfinished degree. The skills stayed. The silence did the harm.
Read
Lesson4 min
What compound annual growth actually measures
CAGR is the constant rate that matches a jagged path's ending wealth. It compresses that history, and it does not forecast what the next year will do.
Read
Lesson4 min
Expected tosses until three heads in a row
First-step analysis gives the mean wait for three heads as 1/p plus 1/p squared plus 1/p cubed. A fair coin gives 14, and 10,000 trials averaged about 13.87.
Read
Method13 min
How many touches before a breakout
One breakout model waits for three touches of a level inside a range that formed after an impulse, and skips the trade when the range is still being built.
Read
Lesson4 min
Why a passion is built by trying things
The path runs from coding to math to markets, including work that was shut down. Passion here means choosing that work, including the part of the time you dislike.
Read
Lesson4 min
Refresh Flask page data without a reload
Ajax can replace one HTML fragment on a timer by posting to a Flask route. The lesson redraws a random number every second without reloading the whole page.
Read
Lesson4 min
Alpha versus simply riding the market
Two simulated funds can share a Sharpe ratio near 2.5 while one is mostly market exposure and the other is not. Alpha is the piece left after that exposure is removed.
Read
Lesson4 min
Why win rate without expectancy misleads
A high hit rate says nothing until you know the average win, the average loss, and the capital behind them. Sales calls do not fill in those missing numbers.
Read
Lesson4 min
Why harsh comments outweigh a quiet room
A bias toward bad news, and a habit of checking for danger, lets harsh comments outweigh silence. The lecture splits that habit into a private fear and a social one.
Read
Method16 min
How to find a daily trading bias
A daily trading bias is the direction you allow for one session, read from the last daily candle and the last closed four-hour candle before you look for a zone.
Read
Lesson4 min
Diversify firm and sector risk, keep market risk
Thirty simulated stocks had 7 percent drift and 20 percent volatility. Equal weight kept market risk and washed out single-name luck. A down market lost about 7 percent.
Read
Lesson4 min
Derive the Black-Scholes-Merton equation
Hedging a call with delta shares removes the random term under geometric Brownian motion. The lesson sets that riskless change equal to the riskless return.
Read
Lesson4 min
Simulate fractional Brownian motion with Davies-Harte
Fractional Brownian motion sums fractional Gaussian noise. Davies-Harte builds the noise with a Fourier embedding. At 64 steps the variance error was about 10 percent.
Read
Lesson8 min
How a failed reclaim decides the Nasdaq open
Chris Creamer says the open is planned as bullish, bearish, or rotational before the bell. A failed reclaim, not a wick, is what confirms the hold. The session's results are his tape, not an audit.
Read
Method16 min
Why this scalp exits at 1.5 times risk
This scalp takes orders from one side of a range and delivers them to the other, then stops near 1.5 times the risk. Most of the lookalike candles in a day are skipped.
Read
Lesson3 min
Why prop-firm challenges resemble a casino
A simulated challenge can collect a fee and never risk firm capital. One 50/50 sketch took $8.9 million in fees and netted about $2.72 million.
Read
Lesson4 min
Gambler's ruin for a trading system
Gambler's ruin is the chance of going broke before a wealth target. With moving win rates and payoffs, a trading system's chance is only an estimate from past trades.
Read
Lesson4 min
Option Greeks as local linear approximations
Delta, theta, vega, and rho are slopes of a Black-Scholes price. Near current inputs, a one-point stock move matched a delta near 0.64. Farther away, the line missed.
Read
Method4 min
How to read a backtest (and when to walk away)
A backtest shows how a set of rules would have behaved in the past. Seven checks separate one you can learn from, from one that was built to be shown to you.
Read
Lesson4 min
What a 2025 drawdown says about sizing
The lecture reports about 44 percent versus 16 percent for the index, a 30 percent hole from short volatility, and a month-long window to redeploy into the rebound.
Read
Lesson3 min
How Fama-MacBeth tests a return factor
Fama-MacBeth regresses next month's returns across stocks, then averages those monthly slopes. In one beta test the slopes were insignificant.
Read
Lesson4 min
Building a live implied volatility surface
A Python screen pulls SPY option quotes, keeps strikes within 2 percent of spot, and draws implied volatility by strike and expiry. A lock button freezes the picture.
Read
Prop firms12 min
How prop firms choose who to fund
A prop firm funds a trader after the sample, the risk, and the payoff can be checked. A described method with none of those three is not treated as an edge.
Read
Lesson4 min
Finite differences for an option price
Finite differences turn a pricing differential equation into a recurrence. Start from a boundary value, step with a small increment, and the grid fills in the price.
Read
Lesson4 min
What a trading sentiment feed is not
The lecture posts a $113,000 Nvidia book and a short put closed for $440 next to a tweet-based score. That score is a reading of the stream, not a buy or sell by itself.
Read
Lesson4 min
Connect Python to the Interactive Brokers API
An Interactive Brokers session in Python pairs a request client with a callback wrapper. A paper-trading login on the local gateway is enough to prove the link.
Read
Lesson5 min
How one gold account scalps the evening reopen
The firm reads one funded gold account that waited for the six p.m. Eastern reopen and scalped inside about two minutes. The host's log math is abnormal, his own word, and it is not an audit or a plan to copy.
Read
Method13 min
Nasdaq order flow scalping strategy
This Nasdaq scalp reads effort in the order flow: aggressive buying or selling that price does not accept, then a stop trailed as that effort continues.
Read
Lesson4 min
How the Kelly fraction sizes a bet
Kelly sizing bets a fraction of the bankroll from the win rate and the payoff per unit risked. A noisy estimate of those inputs can wipe out the improvement.
Read
Lesson4 min
How total expectation splits an average
An average equals each slice's average times that slice's probability. The identity holds when the slices drift. In markets, the lecture says, they drift quickly.
Read
Lesson4 min
Fit a line with the matrix least-squares formula
Least squares picks the intercept and slope that minimize squared errors. In matrix form the coefficients are the inverse of X transpose X, times X transpose y.
Read
Lesson4 min
How a European option payoff actually works
A European call pays the gain above the strike at expiry, in 100-share lots, and only if that gain is positive. Premium shifts the picture, and delta is not a frequency.
Read
Lesson3 min
Building a live volatility regime feed
A live regime screen records five-second bars from a broker feed and refits on the last 60 bars. The first build is the data shell. The classifier comes next.
Read
Method4 min
How a positive-payoff scalp is managed
Matthias shows a Nasdaq scalp with a nine-point stop and about fourteen points of gain. He wants the payoff above one-to-one and stops after two losses.
Read
Lesson4 min
Simulating memory with a Volterra kernel
A Volterra process builds a Gaussian path from past shocks only. The kernel sets the covariance, so one scheme can mimic Brownian motion, mean reversion, or a rough path.
Read
Lesson4 min
What Black-Scholes is actually pricing
Black-Scholes returns a discounted average payoff, not a forecast. Buying the sample call at 14.10 versus a 14.23 model value still lost on single paths.
Read
Lesson4 min
Why generic money advice breaks down
Advice quality varies, and a plan you follow can beat a better plan you ignore. A language model mimics experience without the conditions behind it.
Read
Lesson4 min
Why stock picking can lose to the index
A hot stock can post a higher average growth rate and still leave worse crash paths, because most of the gain is beta plus a volatility penalty.
Read
Method21 min
Liquidity sweep and retest strategy
A liquidity sweep runs price through a level where stops sit, then fails. The retest is the second touch of that level, traded only if the larger structure agrees.
Read
Lesson4 min
How market-neutral portfolios isolate one bet
A market-neutral hedge removes the shared market factor so one long-short bet can win or lose on its own. Beta drifts, so the hedge is reset.
Read
Lesson4 min
How gradient descent, noise, and annealing learn
Training walks downhill on an error surface. Plain steps stop in the nearest valley. Noise, or a cooling rule that sometimes steps uphill, can search for a deeper valley.
Read
Lesson4 min
When a sample average has nothing to settle on
A fair game that pays two to the power of the first tails has infinite expected payout. Repeated averages wander, so the usual limit theorems do not apply.
Read
Lesson4 min
Pricing a variance swap from listed options
A variance swap pays realized variance minus a strike, in cash at expiry. The fair strike is a risk-neutral value replicated by a strip of puts and calls.
Read
Method4 min
Gold scalping strategy at the reopen
One gold scalp waits for the futures reopen near 6 p.m. New York and is usually done within about two minutes. The window is the strategy, not a signal that works all day.
Read
Lesson4 min
What PCA finds in a handful of stock returns
Four stock-return series from 2021-2022 gave a first component near 55 percent that looked like the market. Three components cleared 90 percent.
Read
Lesson4 min
Link StockTwits tone to analyst buy-sell spreads
On rating day, StockTwits tone and the buy-minus-sell spread sloped up together. The fit was statistically detectable and still weak. It is a lead, not a cause.
Read
Lesson4 min
Luck, skill, and a zero-edge book
A zero-edge day-trading book still produces big winners. In one run, 7 of 100 gained over 50 percent. A positive-edge book left 11 of 100 down.
Read
Lesson4 min
Which math courses quants use every day
Algebra, multivariable calculus, and linear algebra show up in daily quant work. Real analysis, abstract algebra, and school geometry almost never do.
Read
Lesson4 min
Trading is a changing game, not a fixed lesson
A fair dollar coin has expected wealth of zero. Once the heads chance wanders, a 30-flip estimate can skip bad stretches. The same estimate on both sides stops the trade.
Read
Method21 min
What risk reward ratio scalpers use
Risk reward is the distance to the exit divided by the distance to the stop. One scalp in this note uses about nine points of risk for about fourteen points of target.
Read
Lesson4 min
What an edge is in a simple market
An edge is a positive expected gain from the action you choose. A dice market has a fixed level of 3.5. Live markets move, and one path can still go broke.
Read
Lesson4 min
Apply Ito's lemma to an option portfolio
Ito's lemma gives the change in an option value that depends on time and a random price. A fixed contract count scales that change into the portfolio.
Read
Lesson4 min
How path signatures price exotic options
Listed options pin down the ending distribution, not the path taken. A signature compresses that path so an exotic payoff becomes a short weighted sum.
Read
Lesson4 min
Counting rare events in markets
A Poisson model counts rare events at a fixed rate, with exponential waits between them. Markets cluster, so the rate is allowed to move with time and with recent shocks.
Read
Method12 min
Inside bar breakout strategy
An inside bar sits entirely inside the prior bar. This breakout does not buy the first poke out. It waits for a retest of the supply or demand the break just left behind.
Read
Lesson4 min
How two closed candles set a scalper's session bias
Harry Sharp's session bias, as told to the firm, uses two finished candles, a clean path, and a spring. The win rate he quotes belongs to his tracker, not an audit.
Read
Lesson4 min
What an early quant offer actually paid
One junior quant offer paid a $156,000 salary plus a $24,000 target bonus, near $180,000 before tuition and signing money. The internship had been $50 an hour.
Read
Lesson4 min
Why portfolio weights drift without rebalancing
A 50-50 Amazon and Apple book bought in 2020 and never reset ends near 40 percent Amazon and 60 percent Apple. The one-time value is about 18,552 from 10,000.
Read
Lesson6 min
What a dull tape did to four prop-session plans
The firm put four traders in one Dubai session under prop rules. Their dollar figures contradict each other. The only steady claim is that the beginner finished green and called it gambling.
Read
Prop firms4 min
Price a prop attempt before the first fee
A worked the firm simulator used 2-to-1 and a 40 percent win rate on a 100k test. At $350 risk, 5,000 runs showed a 65.1 percent pass rate.
Read
Lesson4 min
Why Ito's lemma keeps a second-order term
Ordinary calculus drops higher powers of a tiny step. Ito's lemma keeps one of them, because the square of a Brownian increment acts like elapsed time.
Read
Lesson4 min
How a covered call changes a short call
A covered call is long shares plus a short call. You keep the premium if you hold the short, assignment caps the upside, and a drop through your cost is still a loss.
Read
Lesson4 min
What a hundred million tweets can show a study
A Russell 3000 tweet panel of about 100 million posts shows longer messages over time, a duplicate share near 15 percent, and about 55 to 60 percent naming one ticker.
Read
Lesson4 min
Build a delta-neutral volatility option position
A short call hedged with shares can isolate a volatility view from the stock's direction. Delta sets the share count, and that hedge must be reset as the spot moves.
Read
Lesson4 min
Density of the minimum of three uniform draws
When three independent uniform returns sit on zero to one, the minimum has distribution function 1 minus (1 minus z) cubed, and density 3 times (1 minus z) squared.
Read
Method13 min
The one percent rule in day trading
The one percent rule here is a daily loss limit: risk about one percent, then turn the computer off. One trade is allowed to run. A second idea does not get a fresh budget.
Read
Lesson4 min
How to read a listed options chain
A chain lists calls and puts for one expiry, marks in-the-money strikes, and shows bid, ask, volume, and model Greeks. Delta is a model reading, not a long-run hit rate.
Read
Lesson4 min
What an Ito integral actually computes
An Ito integral is a random variable, not one area under a curve. It adds a weight times each Brownian increment, and a closed form still contains that noise.
Read
Lesson4 min
Pricing is the expected payout of a game
A coin paying 2 dollars on heads has expected payout 1 dollar. Pay 1 and average ending wealth is near zero. A die prices at 3.50. One toy option averaged about 6.70.
Read
Reading claims3 min
How to check a trading creator's claims in five minutes
Most trading content is a screenshot and a story. Five questions turn either into something you can evaluate, and the creators worth following answer all five without being asked.
Read
Lesson4 min
Why a quoted market return is only a model
A cited 10 percent stock return, or 8 percent, is history projected forward. It is not a fact about the next 20 years, and it may skip inflation.
Read
Lesson3 min
Moments of a normal law from one function
A normal moment generating function is an exponential. Derivatives at zero are the moments. Brownian motion then has mean zero and variance equal to elapsed time.
Read
Lesson8 min
Why the entry is the last slice of a Nasdaq trade
Chris Creamer says the click is the last five to ten percent. Location is discount outside value, and confirmation is effort that fails. The July return and the win rate are his account, not an audit.
Read
Method16 min
Market structure, order flow, and liquidity
One breakout model will not separate three reads: the swing structure, the order flow at the level, and the liquidity that has to be taken before price can leave.
Read
Lesson4 min
What Tetlock's pessimism factor found in the news
A pessimism score from one news column lined up with 8.1 basis points of next-day index return. The 7.3 percent strategy did not survive modest trading costs.
Read
Lesson4 min
How retail books differ from trading desks
Retail trading lives one path in liquid markets. A sell-side desk quotes both sides. A buy-side book raises capital for a signal that shrinks as that capital grows.
Read
Lesson4 min
How a C++ pointer shares one memory slot
A C++ pointer holds an address, not a second copy of the integer. Reading it shows the value. Writing through it changes the original, because both names share one slot.
Read
Lesson4 min
European call and put payoffs, long and short
A European call pays at expiry only when the stock beats the strike. Puts reverse that gap. The premium is paid up front and changes gain, not the payoff.
Read
Method4 min
Three checks before a Nasdaq breakout click
Jonas Bearthrop joins Nasdaq breakouts only after structure, the heatmap, and absorption agree. A structure-only test showed about 45 to 50 percent at two-to-one.
Read
Lesson4 min
How lower correlation raises a Sharpe ratio
Lower correlation cuts portfolio variance and can raise the Sharpe ratio if expected return holds. A crisis can erase a zero that was only statistical.
Read
Lesson4 min
How a desk prices a barrier option
A barrier option pays only if the path misses a level. Price it with a model fitted to listed options, then adjust the raw average for risk, liquidity, and margin.
Read
Lesson4 min
Three trading myths checked with expected value
A 50.5 percent win rate can blow up while a 30 percent win rate stays stable. Roulette odds are fixed. Market odds are not, and no rule stays optimal.
Read
Lesson4 min
Why buying the index still needs a liquidity plan
Buying the index for 30 years still fails if job, house, and stocks fall together. A hedge that pays in the drop is the lecture's alternative to selling with the crowd.
Read
Method9 min
How to trade forex from fundamentals
A fundamentals-first forex process scores the economy, the central bank, and the risk mood before it looks for a zone on the chart to time the idea.
Read
Lesson4 min
See the law of large numbers in a simulation
Averages of uniform, normal, or Poisson draws lock onto the true mean as samples grow, and those averages form a bell. That is why a text sample can still be tested.
Read
Lesson4 min
Why a trading bot still has a human bias
A trading bot stores a human choice of model and settings. How fast those settings change decides the year, and judgment stays inside the update.
Read
Lesson4 min
Three calculus results from one practice set
An integral up to x squared has derivative 2 times x times sin of x squared. The area between the parabola and the line y = 4 is 32/3. A related-rate spill grows at 10 pi.
Read
Lesson4 min
Why the derivative definition is useful numerically
The derivative is the limit of a rise over a small step. With a step of 0.0001, the cube x cubed at 1 gives 3.0003, against the exact slope 3.
Read
Lesson4 min
Why the reward follows the work itself
Wanting the prize instead of the work is the lecture's reason the prize stays out of reach. Wealth, here, follows hard problems done for a long time, not a finish line.
Read
Method14 min
Why market context comes before the entry
Context is the higher-timeframe story that decides which side of a level you are allowed to trade. The entry is the last step, and some days the context says do nothing.
Read
Lesson4 min
What a probability model is actually for
A probability model lists outcomes and weights so you can compute an expectation. It does not name the next flip, and it misses any outcome you left out.
Read
Lesson4 min
Derive the Bachelier call price from the forward
The Bachelier forward is a driftless Brownian move, and it is a martingale. The call price is the expected positive payoff, reduced to normal probabilities.
Read
Lesson3 min
Build a Brownian path from normal draws
A Brownian path is a running sum of normal steps. Over 252 unit days, many paths end near mean zero and variance 252. An array build is the practical way to make them.
Read
Lesson4 min
Implied volatility versus what the market later does
Implied volatility is backed out of option prices. Realized volatility is estimated from returns, and a regression of the two is how the lecture shows the premium.
Read
Method6 min
What the Hurst exponent is used for in trading
Traders use a Hurst-style reading to tell a trending tape from a choppy one, then look for a small pullback and a break of the last few bars in the trend's direction.
Read
Lesson4 min
Intermediate quant projects worth building
Build a market-making sim that can use the wrong model, a derivatives pricer you can solve three ways, and an order book with price-time priority.
Read
Lesson4 min
Why stock diversification fails in a crash
Caterpillar and Nvidia looked unrelated in early 2025, then fell with the index. A long put is the lecture's way to break that shared crash and free cash to buy.
Read
Lesson4 min
How a Kalman filter tracks a changing mean
A Kalman filter mixes a fixed mean-reversion model with live prices so the level can drift. The noise setting controls speed, and faster updates bring more whipsaw.
Read
Lesson4 min
Why Bitcoin swings like a new monetary asset
The lecture reads Bitcoin's swings as bottom-up price discovery of a scarce money, rhyming with gold after the 1970s peg broke, amplified by retail attention.
Read
Prop firms8 min
What a prop firm funding contest tests
A funding contest in this session tests whether a trader can explain one setup and then mark a target and a stop on a one-minute chart, under a clock.
Read
Lesson4 min
What deep hedging changes about delta hedging
Fair-price delta hedging targets zero average gain. Deep hedging learns a discrete rehedge from risk-adjusted results, and model risk stays.
Read
Lesson4 min
How a wheel screen picks a strike
The wheel sells calls against shares and puts against cash. Sell the call only when the strike clears your cost. A 30 delta is not the same sale after the price falls.
Read
Lesson4 min
Hedge delta, gamma, and vega with options
Shares can flatten delta, but gamma and vega need other options. A short Nvidia call book is offset with two listed calls, then about 46 shares clear the leftover delta.
Read
Lesson4 min
Price a call with a one-step binomial tree
A one-step tree prices a call by holding delta shares so both branches match. The lesson's numbers give delta of one third and a call price of about 4.70.
Read
Lesson4 min
Which retail trading mistakes matter most
One lecture ranks account-wrecking habits from treating a signal as certain to trusting a gain screenshot. The costly ones skip statistics and chase a story.
Read
Prop firms4 min
What a firm checks before it funds you
The firm says nine in ten traders fail most challenges because a profitable stretch is not yet a process the firm can fund. Survival is the test.
Read
Lesson4 min
How hedge funds stack orthogonal bets
Funds split teams so each return is meant to be uncorrelated with the market and with the other teams. One hedged mix showed a Sharpe near 1.9.
Read
Lesson4 min
Two questions quantitative finance tries to answer
Quantitative finance asks what a contract is worth and which choice is better. A fair charge leaves a hedged book near zero. One path can still crown the worse forecast.
Read
Lesson4 min
Simulate the Bachelier model in Python
A Bachelier path adds volatility times a normal shock times the square root of the step. The lesson draws 100 of those driftless paths across one year with 252 steps.
Read
Lesson4 min
How text becomes an equity trading signal
News and posts can be turned into scores, linked to a company, and sorted into a long-short equity book. The hard part is noise, negation, and a pattern that may fade.
Read
Prop firms5 min
What a prop firm live session shows
A filmed prop session shows how four styles behave under one rule set and a clock. It is a contest, not a method you can copy from the result alone.
Read
Lesson4 min
Plot price paths in a Streamlit app
Streamlit sliders can drive a qfin geometric Brownian motion simulator and redraw a line chart. The lesson's price slider runs from 0 to 100 and starts at 50.
Read
Lesson4 min
Pick a one-off workshop or a project deep dive
Use one workshop for a single technique. Use a deep dive when you can name a public repo, such as a trading system or qfin, and the behavior you want to change.
Read
Lesson4 min
Practice market making with a three-dice game
Three dice hide part of a total while you buy, sell, or quote a spread. The rounds show how a fair price, a one-sided fill, and a balanced book each change the result.
Read
Lesson4 min
Diversification and hedging are different tools
Diversification cuts volatility when assets are not perfectly correlated. A hedge trades an uncertain future for a payoff the contract itself specifies.
Read
Prop firms5 min
Prop firm capital versus your own money
The comparison that matters is gain relative to the cash you personally have at risk: a challenge fee on one side, the whole account on the other.
Read
Lesson4 min
What daily delta hedging says about option prices
Delta is an option's slope against the stock. Selling a 10.45 call and hedging daily left scattered gains. More frequent hedges pulled the average toward that premium.
Read
Lesson3 min
What implied volatility actually measures
Implied volatility is the input that makes a Black-Scholes price match the market. It looks forward. Realized volatility looks back. The smile runs across strikes.
Read
Lesson4 min
The market risk diversification leaves
Nine stocks in three sectors can track the S&P 500 once firm and industry risk are diluted. Extra tech exposure that beats the index is beta, and betas jump in stress.
Read
Lesson4 min
What the Sharpe ratio leaves out of a path
Two paths can share a Sharpe ratio and still differ in return and in downside. The ratio also says nothing about whether that path was fit to old noise.
Read
Method3 min
The six-sleeve portfolio: how the desk's method is built, and why it stands aside
CopyOrders' method runs six independent sleeves on equal capital, each with its own regime filter, so that no single idea decides the record and standing aside is a position, not a failure.
Read
Lesson4 min
Why a fixed return model breaks
A fixed return histogram can look like an edge and then draw down, because the generating distribution moves. A 99.8 percent game price is a quote, not a chance.
Read
Lesson4 min
Why two modest strategies can share a higher Sharpe
A 0.8 Sharpe and a 0.9 Sharpe became 1.2 at a half-and-half mix in the lecture. The gain needs independence. Two equity books can lose it together in a selloff.
Read
Method4 min
Frame the index day with volume and gamma
Manuel frames an index session from accepted volume, then from gamma. He stops after two losers and caps a trade near 1 percent of the account.
Read
Lesson4 min
Solve uniform and binomial probability three ways
A flat density of 1/8 on 0 to 8 puts probability 1/2 on the stretch from 2 to 6. Two fair dice total 7 or 11 in 8 of 36 outcomes. Eight fair trials use a binomial count.
Read
Lesson4 min
How to structure a quant resume for interviews
A quant resume rarely creates the interview. Once you are in the room, it should list only the projects, tools, and results you want the interviewer to ask about.
Read
Lesson4 min
Reading the March 2024 economy and yield curve
Early March 2024 showed a hot stock tape, firm jobs, inflation nearer 2 percent, and an inverted Treasury curve. The lecture reads that mix as a snapshot, not a signal.
Read
Lesson4 min
See why option delta is only a tangent line
Delta is the slope of a Black-Scholes call against the spot. At spot 85 the lesson's call is about 5.0931 with delta about 0.3476, and the line peels away later.
Read
Lesson4 min
Why a normal model misses market crashes
A normal curve fit to S&P returns puts five-sigma days thousands of years apart. The lecture counts 19 such days in about 25 years, and uses regimes instead.
Read
Prop firms4 min
When prop size beats the cash you put at risk
A $2,500 month on $10,000, in the firm's sketch, needs nearly 3 percent risk a trade. The same target on $100,000 of funding needs about 0.35 percent.
Read
Lesson4 min
Multiply two-digit numbers in your head
Split one factor into tens and ones, multiply, and add. Seven products took 4 minutes 51 seconds. Six were right. One miss came from a rushed partial product.
Read
Lesson4 min
Why a stock's expected return will not settle
A die's average settles at 3.5 because every roll shares one distribution. Nvidia's cumulative mean return does not settle across 2008, 2016, or 2024.
Read
Lesson4 min
When a once-profitable trading rule fades
A rule with win chance 0.55, average win 10, and average loss 8 starts with a positive edge. If the win chance drifts toward 0.4, the equity curve flattens.
Read
Lesson4 min
Compound interest and why timing changes value
Money now can earn interest that money later misses. Compounding more often in the year raises the ending amount, and the continuous limit of that amount stays finite.
Read
Prop firms4 min
Ask about drawdown, fees, and payouts first
Intraday trailing can fail a day that is still up $5,000 if open gain is given back. The firm says its trail updates at the close and its first try is $1.
Read
Lesson4 min
Show Python results in a browser with Flask
Flask serves a small Python web app on port 5000 and renders an HTML file. Jinja drops a Python variable into that page so a browser can show server-side results.
Read
Lesson4 min
How Markov chains solve three-heads coin questions
Four streak states turn coin-run questions into matrix powers. Those powers separate ever seeing three heads, ending on them, and first hitting them at toss n.
Read
Lesson4 min
Self-exciting jumps and volatility clusters
A Hawkes process lets each shock raise the chance of the next, then fade. It clusters jumps and cuts a three-sigma wait from about 370 days to the mid-30s.
Read
Lesson4 min
Executive pay and 2021 prices in one cross-section
In one 2021 large-cap cut, top stock awards did not track that year's return. Log awards tilted up with price, and total pay stayed mostly flat across the price range.
Read
Method17 min
Position sizing after the edge is known
Once a tested edge exists, size is a fraction of the edge, not a feeling about the last candle. This account treats the risk model as the thing that carried the result.
Read
Lesson4 min
What three quant interviews showed
Three early interviews pointed to one filter: sit only for the job you want, follow what the interviewer leans into, and do the work before anyone is paying you.
Read
Lesson4 min
Implied probability versus a coin flip
A coin flip's average settles on one half. An implied probability, from odds or option prices, is where the money sits. It is not a frequency from a thousand repeats.
Read
Lesson4 min
Simulate geometric Brownian motion paths in Python
A geometric Brownian motion path steps a price by drift and a volatility-scaled normal shock. This page shows the class, the one-year loop, and the plot.
Read
Lesson4 min
Generate random variables by inverting the CDF
Draw a number from zero to one, then read it backward through the cumulative distribution. That map reproduces a die roll or an exponential draw when the inverse exists.
Read
Lesson4 min
What neural networks actually learn in finance
A neural network fits an expectation, not a future price. That helps approximate a stable pricing function, and it fails when the return distribution keeps moving.
Read
Prop firms4 min
Which style can survive a funded loss limit
The firm calls day trading the cleanest fit under a loss cap, swing trading workable with less size, and multi-week holds a poor fit. Costs still matter.
Read
Lesson4 min
Hidden states behind a return distribution
A single normal fit puts extreme Nvidia moves centuries apart. Three volatility states fatten the tails. A hidden chain learns regimes you do not name.
Read
Lesson3 min
Download daily bars from Interactive Brokers
Daily US stock bars come from the Interactive Brokers API once socket clients are enabled in Trader Workstation. Connect on port 7497 and save each bar callback to a CSV.
Read
Lesson4 min
Time the entry after the liquidity grab
Tim Flossbach, enters only after a liquidation pocket is taken and a turn is visible. He wants at least two-to-one and skips most charts.
Read
Lesson4 min
What a statistical edge actually is
An edge is a price better than the expected outcome, not a candle story. In markets that expectation moves, so a fixed chart rule is not a lasting edge.
Read
Method29 min
How to tell when a trading edge is gone
An edge is gone when the data, not a feeling, says the market no longer matches the model. The fix in this account is to stop that model, not to add a new rule on the spot.
Read
Lesson4 min
The right reading of a 95 percent interval
A 95 percent interval is not a 95 percent chance that the mean lies between two fixed ends. About 95 of 100 intervals built the same way would cover the fixed mean.
Read
Lesson8 min
What a same-day prop contest can actually test
The firm flew five traders to California to compete for five hundred thousand dollars. The host named Brandon after a timed pitch and a half hour. No published gain and loss backs that call.
Read
Lesson4 min
Can a neural net learn an option pricing map
A network can mimic Black-Scholes inside its training box. One row priced at 30.449 came back as 30.679. The useful case is a slow model, such as Heston, learned offline.
Read
Lesson4 min
Why a portfolio can be built without a backtest
A useful portfolio can be designed from how return streams relate, then checked with simple regressions, rather than from a curve that only fit the past.
Read
Method19 min
How to grade a trading edge
An edge gets a grade only after it stays the same long enough to measure. The grade here is asymmetry: how much the winners pay against how much the losers cost.
Read
Lesson9 min
How two ICT Nasdaq scalps set the stop and target
Candace Lao says that her Nasdaq payouts come from two ICT models, with a stated forty percent win rate that needs at least two to one. Those figures are hers and the host's, not an audit.
Read
Lesson4 min
When a trading strategy must stay secret
A strategy that only levers market beta need not be secret. Secrecy starts when you alone hold a dataset, a market-making edge, or an undocumented premium.
Read
Lesson4 min
Read a normal curve by mean and spread
A normal curve is set by its mean and variance. Roughly 68, 95, and 99.7 percent of outcomes fall inside one, two, and three standard deviations.
Read
Lesson4 min
Five projects that opened quant doors
A pricing library, a mortgage chain, a Gaussian cookbook, a long blog, and a volatility system are five projects credited for research and market-making interviews.
Read
Lesson4 min
Classifying live volatility with Markov states
Three volatility states, percentile labels, and a Gaussian update turn five-second bars into a live regime. The fit uses recent bars and can be recomputed on demand.
Read
Prop firms5 min
How to calculate a prop firm challenge
A prop challenge is arithmetic before it is a chart: target, daily loss, max loss, and how many losers in a row still fit. Run those numbers before you pay for an attempt.
Read
Lesson4 min
Solving geometric Brownian motion step by step
Geometric Brownian motion is solved by a logarithm and Ito's lemma. The price is lognormal. A payoff that depends on the whole path still has to be simulated.
Read
Lesson4 min
Draw and sample a normal curve in Python
NumPy can draw standard normal numbers, a short function can trace the bell, and a library call can return the chance of falling below a cutoff.
Read
Lesson4 min
How an AI trading assistant keeps a thesis
An order-placing model needs a saved thesis and hard limits. The demo still dropped data, blocked orders, and moved a target to the last price.
Read
Lesson7 min
What a trader panel required before any funding
Fabio and Andrea, on an the firm pitch day, kept personal money back until a sample could be checked. The offers that survived were smaller prop accounts.
Read
Method20 min
What forced participation means in trading
Forced participation is the moment trapped traders have to exit, and that exit is the move. The entry is the last step, after environment and location.
Read
Lesson4 min
When volatility drag helps or hurts
Half the variance usually taxes compounded growth. A rare high-payoff position can raise geometric growth instead, if you can repeat it without a ruinous loss.
Read
Lesson4 min
Back out implied volatility from a market price
Implied volatility is the Black-Scholes input that matches a traded option price. One Apple call solves near 31.38 percent and reprices the three-dollar quote.
Read
Lesson4 min
How portfolio risk and return actually fit
Risk is not a coupon for higher return. Historical averages do not pick your future path, and diversification inside one market still leaves that market's risk.
Read
Lesson4 min
Why a reached goal feels smaller than it did
Expectations and the band of ordinary days both move as you adapt. A long goal, once reached, already sits inside that band, so the trip carried the weight.
Read
Method4 min
Backtest vs live trading: where the gap comes from, and how to read across it
A live record almost always underperforms the backtest it came from. The gap is not a scandal; it is five specific things, and a record that keeps the labels apart lets you measure it.
Read
Lesson4 min
Common mistakes in quant comment threads
A comment thread mixes backtest habits, a false résumé, and a 93 percent win rate. The replies separate political risk, sample size, and the size of the losses.
Read
Lesson4 min
Why one winning trade is not an edge
One earnings gap has no other version. Edge here is a positive average trade whose inputs keep changing, so a chart line or a long backtest cannot settle it.
Read
Method4 min
Name the day's context before any entry
Tom Vorwalt marks balance, the prior close, and liquidity before any setup. A 40 percent hit rate, he says, needs more than 2-to-one to earn its keep.
Read
Lesson4 min
Backtest a moving-average crossover in pandas
A pandas backtest of SPY goes long when the 7-day average exceeds the 14-day average and short otherwise. The lesson's Sharpe for that test prints near 0.697.
Read
Lesson4 min
Approximate an integral with random samples
Monte Carlo integration averages a function on uniform draws and multiplies by the interval length. For 4x cubed from 0 to 2, 10,000 draws give about 15.985 versus 16.
Read
Lesson4 min
How a levered long-short book breaks
A short borrows shares and can be forced shut. Funding longs with that borrow is leverage, and the lecture cites a 42 percent weekly tail on a ruined book.
Read
Lesson4 min
Repricing a straddle after volatility falls
A short straddle gains if implied volatility falls and the shares stay put. On one Nvidia chain, a drop to 20 percent volatility showed a gain a 10-point slide erased.
Read
Method11 min
Why the trade comes after the news print
Trading the news, in this account, starts after the number is out. The first half hour sets the size of the battle. The middle of a noisy week is left alone.
Read
Lesson3 min
What the qfin Python library can price
qfin prices European calls and puts with Black-Scholes and runs Monte Carlo on several exotic payoffs. Geometric Brownian motion or the Heston model drives the paths.
Read
Lesson4 min
Run Python work on several threads at once
Python threads let several waits run together. A two-second task done four times in a line takes four waits; started on four threads, it finishes in one.
Read
Lesson4 min
Probability problems from sample means to Ito
On five log returns the mean, median, and mode can all equal 0.01. The same ladder ends at geometric Brownian motion and a rough-volatility question.
Read
Lesson4 min
How a backup plan changes the odds you quit
The lecture draws more chance of quitting while a backup remains, and more chance of finishing once it is gone. The pictures are qualitative. They are not measured odds.
Read
Lesson4 min
How random variables map outcomes to numbers
A random variable turns each outcome into a number. Discrete laws use a mass function. Continuous laws use a density, and probability is an area.
Read
Prop firms6 min
Which trading style fits prop firm rules
A style fits a prop account when its normal losing streak stays inside the daily and overall loss limits. The style that earns the most on a personal account may not.
Read
Lesson4 min
Normal return odds, portfolio mean, annuity due
Mean 7 percent and standard deviation 2 percent put a return below 5 percent at probability 15.87 percent. Three weights give 9.3 percent. An annuity due ends at 1,195.
Read
Lesson4 min
Cache live option ticks for later analysis
Live option ticks disappear unless the callback stores them. Three lists on the wrapper keep delta, the option price, and the underlying price when a field changes.
Read
Lesson4 min
How five prop styles ranked on challenge data
On the firm's book of about 10,000 traders, order flow led and reversal finished last. Stated pass rates ran from 5 percent to about 15 percent.
Read
Lesson8 min
When a small range breaks and a large one fades
Patrick Nil says that about seventy percent of days are ranges, and that a small zone is a breakout while a large one is a fade. The World Cup record and win rates are his and the host's, not an audit.
Read
Method22 min
What is a liquidity grab in trading
A liquidity grab is a push through obvious stops that fails to continue. The trade, if any, waits for the reversal after the grab, and most setups are skipped.
Read
Lesson4 min
What to do after a run of bad outcomes
After a cluster of bad events, the lecture judges the next action, not the stretch. Repeating a harmful action is framed as taking from your future self.
Read
Lesson9 min
Why a low win rate can still be the whole plan
Pietro Velasro says his World Cup results came from a method that wins about thirty-two to thirty-five percent of the time. The rate, the places, and the payoff rules are his account, not an audit.
Read
Lesson4 min
Why a swap fails until you pass by reference
Copies passed into a function cannot swap the caller's numbers. A pointer or a C++ reference reaches the originals, and the same three assignments then exchange them.
Read
Lesson4 min
Linear independence and what it does to a model
Vectors are linearly independent when only zero weights sum them to zero. Dependent features repeat each other and cannot cover the space a model needs.
Read
Prop firms5 min
Why prop firm accounts fail
Most failed prop accounts in this firm's review did not die from the entry. They died from one behavior: adding size, moving a stop, or trading after the daily limit was already gone.
Read
Lesson4 min
Beginner projects that teach quant work
Start with a performance dashboard, a grid-search visualizer, and a binomial option tree. Each project teaches a job skill that a generated repo will not.
Read
Lesson4 min
Casino edge and a market maker's spread
Roulette, craps, and slots pay the house about 5.26, 1.4, and 2 to 15 percent per play. A bank's analogue is a hedged bid-ask spread, which is still not a locked gain.
Read
Lesson4 min
Why one backtest mixes different trades
The same entry at two times can draw from different gain distributions if history matters. Pooling those trades is why many backtests misread a strategy.
Read
Lesson4 min
Read a linear subscriber growth chart
A subscriber series in this session looks nearly straight, and the average daily gain is about 1.88. A two-point slope check is about 2. No 2023 total is fitted.
Read
Lesson4 min
How the Black-Scholes equation is built
Hedge the option with delta shares and the random terms cancel. The hedged book must earn the riskless rate, which produces the pricing equation.
Read
Method16 min
How seasonality is tested in trading
A seasonal trade is a calendar window that rose in most past years, entered at the open of the window and exited at the close, then checked on years the rules never saw.
Read
Lesson3 min
Five Python habits that speed up quant code
Slice notation, string splits, column filters, least squares, and saved objects cover most daily quant Python. Here is what each does and where it breaks.
Read
Lesson4 min
Why bet size changes a wealth path
A positive average does not keep a proportional bet safe. Most paths go broke until the stake follows Kelly, and half of that fraction is the cautious version.
Read
Lesson4 min
Pricing options when no formula exists
If a pricing equation has no closed form, the same no-arbitrage price is an expectation you can average by simulation. The paths must come from the model you claim.
Read
Lesson3 min
Why a sum of money today differs from later
A sum today and a sum later differ because interest connects them across time. This first session frames present value, continuous time, and a spreadsheet comparison.
Read
Method5 min
What is an NR4 breakout strategy
An NR4 is the narrowest range of the last four bars. The breakout trades the expansion out of that quiet bar, with the stop on the other side of the range.
Read
Lesson6 min
How an adapted NR4 waits for volatility to shrink
The firm reconstructs an unnamed funded trader whose breakout waits for four shrinking hourly ranges, then one break. The payout band and the payoff ratio are the host's reading of one account, not a forecast.
Read
Lesson4 min
Estimate a probability with Monte Carlo in Python
Monte Carlo estimates a probability by repeating a random trial and averaging the successes. A 70 percent shooter and 10,000 rounds of 100 shots show the method.
Read
Lesson4 min
What ARCH and GARCH capture in returns
ARCH lets variance depend on recent squared shocks, so tails fatten and volatility clusters. GARCH adds lagged variance. A short GARCH can stand in for a very long ARCH.
Read
Lesson8 min
How three session VWAPs time a one-minute scalp
Robert Rother says he fades or joins the first touch of a session VWAP, with the book as confirmation. The one-minute fill he shows is two hundred seventy-five dollars on the account on screen. His history is the host's introduction.
Read
Method14 min
What gamma exposure means for index trading
Gamma exposure is how options positions can pin or release an index. This framework starts with where volume accumulated, then asks who is forced to hedge.
Read
Lesson4 min
n-step odds in a two-coin Markov chain
Two coins plus a heads-or-tails switch form a Markov chain. Coin one on day two has probability 0.65. Longer dates use powers of the one-step matrix.
Read
Lesson4 min
How a cash-secured put is paid and how it breaks
You collect a premium for the duty to buy 100 shares at the strike. Puts look rich versus later moves, and a high win rate can still end the account.
Read
Lesson4 min
What AI changes about mathematical research
A model that restates your theorem threatens credit, not curiosity. What remains is choosing a risk and iterating faster than a plan that never ships.
Read
Lesson4 min
How Markovian lifting approximates rough volatility
Rough volatility remembers the whole path, which is slow to simulate. Markovian lifting replaces the power-law kernel with a finite stack of mean-reverting factors.
Read
Lesson4 min
Price a European call with Monte Carlo paths
Monte Carlo prices a European call by averaging discounted payoffs across simulated paths. One worked case uses 1,000 paths and prints 160.75 against a 154.46 quote.
Read
Method17 min
How to trade a pullback to VWAP
A large order often has to work back toward the session's volume-weighted average. The pullback trade is that return, not a guess at the high or the low of the day.
Read
Lesson4 min
What broken model assumptions do to risk numbers
Amazon daily returns in 2021-2022 had kurtosis 6.35, so a normal tail estimate was too thin. Correlation and a moving mean bend risk the other way.
Read
Lesson4 min
Why the first million takes the longest
A teacher saving under $1,000 a month needs about 27 years to reach $1 million at a 7.5 percent real assumption. Later years compound a larger base.
Read
Lesson7 min
How many trades before a win rate means much
Gianluca Bruni says that ten trades can disguise a sixty percent win rate. He wants about a hundred before the rate means much. The session's dollars are not that test.
Read
Reading claims3 min
What a labeled record looks like, with the desk's own as the example
Three kinds of numbers exist in trading research, and almost every record you will ever see blends them. Here is what it looks like when they are kept apart, figure by figure.
Read
Lesson4 min
Why the Sharpe portfolio is not the growth objective
The Sharpe ratio and compounded growth are different goals. A sleeve with negative growth can lift the mix. Leverage without that sleeve can deepen the hole.
Read
Lesson3 min
Why a stock-price model is needed before expiry
Strike and expiry of a European call are known today. The stock at expiry is not, so a price process is how today's information becomes a distribution you can price.
Read
Lesson4 min
How market makers set a fair quote
A market maker quotes around an expected value, half a spread in a simple dice market. For options that value is a model price, and hedges plus a bad model consume it.
Read
Method13 min
Order flow scalping above VWAP
This order-flow scalp only stays with a long while price holds above VWAP. If the idea is already wrong, the stop is cut before the market fills it.
Read
Lesson4 min
What poker and trading both train
Firms host poker because both games reward a policy, not one lucky hand. Roulette's edge is fixed and negative. A poker action can change that expectation.
Read
Lesson3 min
Fair price of a consecutive-heads coin game
Pay D each time heads follows heads in n flips. Expected payout is (n minus 1) times p squared times D. A fair coin and a one-dollar prize pays 0.25 on two flips.
Read
Lesson4 min
Four ways to use a million dollars
The lecture's four sketches are Treasuries, public equities, a private venture, and spending the pile down. Only the first is treated as riskless.
Read
Lesson4 min
Why day-trading courses leave no durable skill
A course that blames you when the pattern fails does not leave a testable skill. Durable skill is probability, position size, and work someone can hire you for.
Read
Method13 min
What is an ICT scalping strategy
An ICT scalp, in this account, is one of two Nasdaq models: a liquidity sweep that fails, or a second leg into a fair value gap after structure shifts.
Read
Lesson4 min
Learning quant finance without a target school
A quant path can start with weak high-school math and a non-target campus if the work is pricing, probability, and projects you can explain in an interview.
Read
Lesson4 min
Why fear misprices ordinary personal risks
People put too much probability on rare disasters, so ordinary actions look like bad bets. Someone who keeps taking the action is a sign the tail was priced too heavily.
Read
Lesson4 min
Use a sample mean when the distribution is unknown
A hypothesis test can ask which side of a field holds the ball without knowing the raw distribution. A large sample makes the average nearly normal.
Read
Lesson4 min
Using Black-Scholes for relative option prices
Black-Scholes turns two premiums into comparable implied volatilities and deltas. Both describe the market now. A delta is not a hit rate you can count on later.
Read
Method19 min
What a failed breakout at the open looks like
A failed breakout at the open is a move that cannot hold the level it just took. The trade, when there is one, is the flip after that failure, not the first push.
Read
Lesson4 min
Brownian motion and a terminal price
Brownian motion starts at zero and spreads with time. A European call is the expected discounted payoff under that terminal law. Long samples approach the model price.
Read
Lesson4 min
Solve geometric Brownian motion in closed form
Geometric Brownian motion solves through a log transform and Ito's lemma. The later price equals the start times an exponential of drift and a Brownian move.
Read
Lesson4 min
What a time series forecast actually is
A time-series forecast is an expectation if the recent pattern continues. It is not a call on a break. Filtering estimates the present. Smoothing estimates the past.
Read
Lesson4 min
Why a crash can help a prepared portfolio
A crash hurts people who must sell into it. A portfolio that holds convexity or spare cash in that state can buy assets others are forced to unload.
Read
Lesson4 min
Reading a forward volatility regression
Regress next-month implied volatility on today's level. In one Nvidia window the high-vol fit was stronger, and the live reading sat near the split.
Read
Method9 min
VWAP pullback strategy for a trend day
On a trend day, a VWAP pullback is a return to the session average in the direction of the drift, taken only while that drift is still intact.
Read
Lesson4 min
Three steps toward a quant role
This hiring path is skill first, a conversation that creates the interview second, and the technical round third. Freezing usually means the material is not mastered yet.
Read
Lesson4 min
How to read a market bubble as expectations
A bubble label is a forecast in disguise. The lecture treats prices as expectations that can break, and asks whether your portfolio survives the break.
Read
Lesson4 min
Three biases that fake a strong backtest
Look-ahead bias, parameter snooping, and dropped dead firms can all paint a smooth equity curve. Removing each one shows whether a signal was ever there.
Read
Lesson4 min
How a hedge sleeve changes a stock portfolio
A stock index here shows a 0.77 Sharpe ratio and a 25 percent drawdown. A dull sleeve, then leverage, is the attempt to cut that drag and keep the engine.
Read
Prop firms4 min
How one bad day ends a funded account
The firm's review of more than 5,000 failed accounts says over 90 percent died from one or two trades inside 24 hours. The figures are the firm's.
Read
Lesson4 min
Implied volatility surfaces and consistent quotes
Implied volatility is the input that makes Black-Scholes match a market price. One volatility cannot price every strike and expiry, because real surfaces are not flat.
Read
Lesson4 min
Volatility crush as an insurance sale
Implied volatility often falls after an event, and a short option gains if spot sits still. A large share move can overwhelm that drop, as one Dell straddle showed.
Read
Lesson4 min
How to study quant topics with a language model
Use a language model to map prerequisites, draft notes, and grade gaps. Do not advance with a partial score, and do not treat the model as a source that cannot be wrong.
Read
Lesson8 min
Why a weak payoff can still pass prop challenges
Matteo Conti says a VWAP pullback wins often and pays poorly, which his simulations pass challenges and do not suit a private account. The pass rates are his test, not an audit.
Read
Method16 min
Order flow: effort versus result
Effort versus result in order flow asks whether the trading that showed up actually moved price. A lot of effort and little result is a failed auction, not a reason to join it.
Read
Lesson4 min
How to study quant topics in order
A three-level map runs from coding and algebra up to stochastic calculus. Traders stop earlier in math than researchers. Each level ends with a check.
Read
Lesson4 min
Read call and put payoffs at expiration
At expiration a call pays the positive gap between spot and strike, and a put pays the gap the other way. Two XYZ cases and a 50-50 example show the arithmetic.
Read
Lesson4 min
Why the quoted price of risk is off
Quoted risk is the crowd's breadth, not a true probability. Deeper work can disagree, as when Ackman read mall debt in the financial crisis.
Read
Lesson4 min
How a college student reached a quant desk
A non-target student reached a quant desk by publishing work and asking people to pass a resume, not by blasting applications. The offer arrived before the degree did.
Read
Lesson6 min
What four funded traders use to clear a test
Four the firm traders clear a test with one or two planned trades. One copied a Nasdaq long across ten accounts for about $6,000 in under a minute.
Read
Method12 min
Scalping versus options trading
Scalping and options trading can run on the same morning and still be different businesses: one hunts a short mechanical edge, the other reads options order flow.
Read
Lesson7 min
How a Hurst proxy gates a five-bar breakout
The firm reconstructs a funded trader's fifteen-minute rules: a Hurst proxy above fifty, slight anti-persistence, then a close through the prior five-bar high. The monthly figure is the host's, not an audit.
Read
Lesson4 min
Yesterday's return does not predict today's
A scatter of S&P 500 returns against the previous day is a cloud around zero. A line scored about 0.99 in sample and 33.50 later. A neural net scored 0.952 and 33.75.
Read
Lesson4 min
Why Black-Litterman weights move less than MVO
Mean-variance weights jump when historical returns twitch. Black-Litterman starts from market-implied returns, then lets a stated view pull the mix away from that anchor.
Read
Lesson4 min
Quant signals and discretionary policies
American roulette pays the player -5.26 percent a spin. An election price can read 80 percent and still be one vote. Quant and discretion differ in the actions.
Read
Method4 min
Enter when broken supply turns into support
The firm's founder buys a breakout after broken supply holds as support. On gold he took about six points, then left the rest running near 40.
Read
Lesson4 min
What poker shows about a trading backtest
Pocket aces win when nobody changes behavior, then lose when opponents do. Folding big action restored a positive expectation by cutting the average loss.
Read
Lesson4 min
Test same-day sentiment against same-day returns
Same-day StockTwits tone and same-day returns lined up across daily quartiles. One long-short book in the lecture had a Sharpe near 2.3, and it was not tradable.
Read
Lesson4 min
Compare buy-side P quants and sell-side Q quants
Q quants calibrate continuous models to today's listed prices. P quants forecast with large discrete data sets. A rich option versus a model price sits on the border.
Read
Lesson4 min
Reading an earnings volatility crush
One Nvidia earnings date cut implied volatility from 43.5 to 32.1 percent while the shares barely moved. A short-volatility position would have gained that day.
Read
Prop firms5 min
What is a prop trading firm
A prop trading firm rents you access to a larger balance after an evaluation. The fee, the intraday loss rule, and the payout terms decide whether that access is usable.
Read
Lesson4 min
Why coin-flip math fails on returns
A two-for-one coin has an edge only while the odds stay fixed. Regimes move, and one 2025 sample mean sat 1.7 million years out on a normal curve.
Read
Lesson4 min
Run two Python market checks without blocking
A price check that loops forever blocks later code. Threads run two checks on separate clocks, and a flag stops them without restarting the program.
Read
Lesson4 min
Pricing Heston options by Fourier inversion
Carr and Madan damp a call so its Fourier transform exists, swap in the Heston characteristic function, and invert. The price comes back without a new Monte Carlo run.
Read
Lesson4 min
What alpha means in a pricing model
Alpha is the intercept after you account for a pricing model's factors. Beating the S&P 500 can be beta, and the intercept can vanish if you add factors.
Read
Method4 min
What drawdown is, and why it is the number that decides whether you keep going
Drawdown is the fall from a peak to the trough that follows. It tells you what a strategy would have asked you to sit through, which matters more than the return it would have paid.
Read
Lesson3 min
How to separate alpha from market beta
A rising equity curve can still be market beta. Compare the strategy with the market over the same dates, including a selloff, before treating leftover return as alpha.
Read
Lesson4 min
What the open Gaussian cookbook contains
The open cookbook simulates Brownian motion, a bridge, fractional paths, and a Markovian lift. Extra factors track covariance. The lift is faster and still approximate.
Read
Method14 min
Value area trading strategy
Value area trading sells from one area of accepted price toward the next. This session refuses the open, refuses a chase, and only sells when value itself has shifted.
Read
Lesson4 min
Why Monte Carlo prices match Black-Scholes
Sample averages approach the expectation as draws accumulate. Success chance 0.7 averaged 0.711. An option simulation printed 10.4504 beside a formula price of 10.45.
Read
Lesson4 min
Learn Python by building one stock chart
Learn Python by finishing one interactive stock app: charts, indicators, and a date window. Beginners start at the base; people who already code skip ahead.
Read
Lesson4 min
Why mean-variance weights fail out of sample
On three games with fixed odds, an optimized mix lifted a Sharpe ratio from 1.83 to about 1.92. The same fit on stocks chases a history that will not repeat.
Read
Lesson4 min
Arithmetic versus geometric Brownian motion
Arithmetic Brownian motion adds shocks and can go negative. Geometric motion scales with price, stays positive, and drops half the variance from its drift.
Read
Lesson3 min
Why classroom statistics break on live prices
Textbook estimates converge only if the distribution stays fixed. Live returns mix regimes, so the mean is noisy and the real choice is model risk.
Read
Prop firms13 min
Three strategies for a prop firm challenge
Three systematic ideas for one challenge: a break out of a noise band, a VWAP pullback armed by the opening range, and a single overnight trade aimed at three to one.
Read
Lesson4 min
Speed up rough-volatility calibration with a network
Rough-volatility prices are slow to simulate. A small network can learn the map first, and a barrier-option check in the paper stayed under 10 basis points of error.
Read
Lesson4 min
Why a unit root fools a price regression
Regressing Apple's next close on today's close gives an R-squared near 0.998. The same fit on returns falls to about 0.015 because the price has a unit root.
Read
Lesson4 min
Why a trading edge is hard to estimate
Edge multiplies typical wins and losses by how often each occurs. Those inputs move, and a short sample estimates them badly. A positive number is not a gain promise.
Read
Lesson4 min
What a quant job is actually like
A quant seat is mostly data work that often leads nowhere, pays less than it looks, and takes years of math before the title politics even start.
Read
Prop firms4 min
What 1 to 3 percent pays on 200k funding
On a 200k funded balance, the firm's host puts 1 percent near $2,000 a month and 3 percent near $6,000. Fees and failed attempts still count.
Read
Lesson4 min
Why Monte Carlo estimates settle on a mean
Monte Carlo works when every draw comes from one fixed distribution. The sample average then settles on the true mean, which is how a game or an option gets a price.
Read
Lesson4 min
Why a GPU speeds PyTorch matrix math
A huge matrix product took about 23 seconds on a CPU and about 3 seconds on a free GPU, once the clock waited for the device. Training a network speeds up the same way.
Read
Lesson7 min
Scalp the VWAP pullback a large order creates
Robert Rother scalps ES pullbacks at day VWAP with an 11-tick target and a 10-tick stop. He states a win rate near 75 to 80 percent, his tally, not an audit.
Read
Lesson4 min
Will AI shrink quant teams or move the work
A small failure rate times a huge pile of automated tasks creates many edge cases. The lecture expects smaller quant teams, and more people fixing what the models break.
Read
Method4 min
Fade the box that forms after a thrust
Patrick Niel, a repeat world-championship finalist, fades the range after an impulse and caps his own book near 1 percent. He cites a 50 to 60 percent win rate.
Read
Lesson4 min
How volatility drag lowers geometric growth
Geometric growth is about the average return minus half the variance. A positive average can still compound poorly when the path swings hard.
Read
Lesson4 min
How to read a portfolio alpha and beta
Beta is how your returns move with the market, and alpha is the leftover. A rolling regression shows both changing, and neither number is a forecast.
Read
Lesson4 min
Stream live option quotes with the broker API
A live option stream starts with a contract plus a market-data request. The wrapper callback prints the option price and the underlying price as ticks arrive.
Read
Lesson4 min
Which college majors suit quant work
On this ranking, mathematics sits alone at the top if you learn it to use it. Data science is next. Finance and computer science rank lowest for most quant seats.
Read
Lesson4 min
Quant trading is incomplete information, not roulette
Roulette fixes the house edge, and one path can finish flat while the average rises. Poker and trading change the edge by acting. A good average can still lose.
Read
Prop firms5 min
What a 200k prop firm account changes
A larger funded balance changes the dollar value of one percent. It does not remove challenge fees, failed attempts, or the firm's loss limits.
Read
Lesson4 min
Why a trading policy cannot be handed over
Roulette's player edge is about -0.0526 per dollar. Doubling after losses is banned. Poker and trading need a changing policy, and that policy is learned.
Read
Lesson4 min
How to rebuild after a project is destroyed
A collaborator with admin access can erase years of work in one act. Revoke that access, reseed what you still know, and keep the project small on purpose.
Read
Lesson4 min
Connect a Java trading client to Interactive Brokers
A Java bot reaches Interactive Brokers through a gateway, a jar on the classpath, and callbacks. A reader thread holds the socket. Crypto needs API version 10.2 or newer.
Read
Lesson3 min
Why option prices ignore your forecast
Two traders can disagree on the drift and still share one option price. A complete-model hedge removes the drift. Live prices come from the volatility surface.
Read
Method16 min
Can a low win rate trading strategy work
A low win rate can still be a workable strategy when the winners are larger than the losers and the losing streak fits inside the risk limit you actually have.
Read
Lesson8 min
Why a liquidation sweep still needs the retest
Tim says that a sweep without a reclaim is not the trade. The short starts once price falls back through the order block, and the minimum payoff he repeats is two to one. His hit rate is his claim, not an audit.
Read
Lesson4 min
Worked examples from arithmetic to open research
The same person can compute 37 plus 48 and also state an unsolved fluid equation. What changes is the object: a number, then a proof, then a question nobody has closed.
Read
No notes match that search.