Backtests,
honestly.
Plain-language essays on why strategy claims fail, what a backtest can actually prove, and the checks that separate a real edge from a well-dressed accident.
No signals No promises No affiliate links
Why Most Backtests Lie — and the 7 Checks That Catch Them
A beautiful backtest is the easiest thing in trading to fake, usually by accident. The seven ways backtests mislead, and the check that catches each one.
Read the essayWalk-Forward Testing, Explained Honestly
Two different things travel under the name “walk-forward”. Know which one you are actually getting — period stability, or purged and embargoed training.
Read the essayAI Trading Bots in 2026: Real Returns vs Advertised Claims
The dashboards all glow green. Here is what independent checks actually measured, and the five questions that separate a real record from marketing.
Read the essayBefore You Trust a Strategy: How to Audit Someone Else’s Backtest in 10 Minutes
Four questions expose most weak backtests in minutes: trial count, costs, sample span, and what chose the parameters. Plus the baseline test every claim must pass.
Read the essayBacktest vs Live: Why Execution Friction Kills Good-Looking Strategies
Your strategy’s logic was never the problem. Latency, slippage and costs are the gap between a beautiful backtest and a losing live account — and the gap is measurable.
Read the essayPBO, Explained in Plain English: The Probability Your Best Backtest Is a Fluke
You tested 8 strategies and one looks brilliant. PBO asks the question most backtests never face: among everything you tried, how likely is the winner just the luckiest loser?
Read the essayWhy a 70% Win Rate Over 20 Trades Means Nothing
70% sounds like an edge. Over 20 trades, it is statistically indistinguishable from a coin flip. The Wilson interval shows why — with the numbers worked out.
Read the essayBeat Nothing First: The Baseline Test Every Trading Model Must Pass
Before a model earns your money, it must beat doing nothing. The always-bullish and no-change baselines are the humblest — and most skipped — test in trading.
Read the essayMinTRL: How Long a Track Record Do You Actually Need?
A Sharpe ratio of 2.0 on 18 months of data proves almost nothing. Minimum Track Record Length turns “impressive Sharpe” into “how many years would this take to believe” — with the numbers.
Read the essayHow to Audit a Vendor’s Backtest Before You Pay a Single Rupee
You are not the one who built the strategy — you are the one being asked to pay for it. That makes you the auditor. Here is the buyer-side checklist, in order.
Read the essayAI Agents Are Trading Real Money Now — Here Is the Verified Record, Not the Hype
Forget the screenshots. On-chain and venue data from November 2026 shows what AI agents, vaults, and apps actually did with real money — and 17 of 18 advertised claims still can’t be checked.
Read the essay92% Fail FTMO — The Math Behind Prop Firm Failure
FTMO's own data says 92% of traders fail its challenge. The cause isn’t bad strategies — it’s one rule, one morning, and position sizes that make failure a matter of time.
Read the essaySEBI’s FY26 Data: 88% of Retail F&O Traders Lost ₹91,685 Crore
SEBI's August 2026 study is out: 87.7% of individual F&O traders lost money in FY26, options drove 92% of losses, and participation fell 20%. The numbers are brutal — and barely moved.
Read the essayChatGPT Can’t Backtest: Why “AI” Trading Bots With 10-Year Backtests Are Fiction
A bot that needs live ChatGPT calls cannot have a 2019 backtest — ChatGPT didn’t exist. That single fact exposes most “AI EAs” on the market.
Read the essayJane Street Made $39.6 Billion While AI Trading Contests Lost Money
Bloomberg reported Jane Street’s $39.6B trading revenue in 2025. Meanwhile, public contests pitting leading AI models against markets mostly lost money. Both get called “AI trading.” Only one description is honest.
Read the essayGrid Bots and the Annualised APR Trick: How a 3% Fortnight Becomes “300% APR”
A grid bot earned ~3% in two weeks in a sideways market — then stalled when price broke range. The dashboard annualised that fortnight into a triple-digit APR. Here is the trick, step by step.
Read the essayThe Daily Drawdown Trap: Why One Bad Morning Ends Your Prop Challenge
You can be net profitable for two weeks and still fail a prop challenge — because one bad morning breached the daily cap. The rule, the math, and how to survive it.
Read the essayFrom Backtest to Funded Account: The Validation Checklist Prop Traders Need Before Paying
Challenge fees are $200–$1,000+ per attempt, and ~90% of attempts fail. This checklist — run before you pay, not after — is the cheapest money in prop trading.
Read the essayCrypto AI Vaults: 3 of 6 Beat Bitcoin, 1 Fell 30% — What That Actually Proves
The most checkable corner of AI trading is Hyperliquid vaults — public numbers, no screenshots. The verified record is mixed, and the mix itself is the lesson.
Read the essay9 Out of 10 Lose: The Real Odds in Retail Trading, and the Only Edge That Matters
SEBI: 88% of F&O traders lost money. FTMO: 92% fail. AI apps: most users lost. Three different markets, one pattern — and one edge that actually changes the odds.
Read the essayWritten to the same standard as the listing.
No article here promises a return, sells a signal, or describes the harness as more than it is. Walk-forward means period stability, not purged walk-forward. PBO needs a candidates file and is never a gate. Accuracy is always recomputed from raw predictions, never taken from a claimed column.
If an essay says a check exists, you can read exactly what it does and where its boundary is on the eight checks
A backtest should have to prove itself.
That is the whole product: a harness that recomputes the number, beats it against baselines, and returns a verdict with an exit code.