@tradegrub.bsky.socialMost people think migrating to Python is just rewriting syntax. Wrong. The real trap: your old backtests lied to you. Slippage, order fills, timezone bugs, floating point math—they're all hiding in your pine scripts. #Python #Backtesting #TradingStrategy
@tradingstrategies.workV2 - Start a backtest
What does a backtest tell you, and what does it not?
Pick a pair, a rule, a span of years, press run.
Every trade the rule would have made shows up at once.
Step by step, under three minutes.
@tradingstrategies.workWhat is Backtesting Arena?
You've seen the post: a chart with arrows, an indicator that "called every top", no word on how often it missed.
This is where you check it: one rule, one pair, real prices, costs included.
First video of the series.
@tradegrub.bsky.socialOur community just shipped an RSI mean-reversion strategy that trades on real data, not assumptions. When price overshoots below 30, it signals a setup. The logic is tight: entry on bounce, exit on mean return. #TradingStrategies #RSI #Backtesting
@tradingstrategies.workFor years our backtests compared Buy & Hold over one period against the strategy
over another, then called the difference a verdict.
On our reference run that gap was 12.55 points. Enough to flip it from −9.1 to +3.4.
Here is what we changed. 🧵
@allthistrading.bsky.socialI tested 120 fictional trades. The session split changed the result: Asia 0R, London +8R, New York -3R. Tag time before trusting the headline. Visit https://join.att-vip.com/ #TradingEducation #Backtesting
@tradingstrategies.work1/9
You subscribe to a trading algorithm and get a number: +27 % a year.
What you do not get is the number that decides whether you are still in it in year three: how far the account fell, and for how long.
Even the good providers leave it out. 🧵
@tradingstrategies.work9/9
Before believing any "after X came Y": ask for the drift, the pairing, the independent count, the locked definition.
Run any state:
tradingstrategies.work/market/bitco...
Not investment advice, not a forecast.
Study the Past — Improve your Future 🥋
#backtesting
@tradingstrategies.work1/9
Bitcoin made a median +31 % in the six months after quiet volatility.
It made +27 % in the same six months without the condition.
The first number gets posted. The second one decides whether the first means anything.
@tradingstrategies.work8/8
Watch: whether the 20-year hold survives the amendments, and the vote by party.
A committee is not the House. A markup is not an outcome. A custody framework is not a purchase.
tradingstrategies.work
Study the Past — Improve your Future 🥋 #backtesting
@tradegrub.bsky.socialYou load five indicators onto your chart. Backtests look perfect. Live trading? Disaster. The indicators fit the past, not the future. Real edge comes from fewer, stronger signals. Stack deliberately. Test on out-of-sample data. #TradingStrategy #Backtesting #AlgoTrading
@tradegrub.bsky.socialMost backtests are lies. Walk-forward optimization splits your data into expanding windows so you're not training on tomorrow's prices. You build on old data, test on fresh data, repeat. #Trading #Algo #Backtesting
@tradingstrategies.work10/10
Nearer event first: 18 Sep weekly, 18,275 contracts, max pain 78,000, at spot.
A strike with n=1 behind it is not a forecast. OI structure and vol pricing are readable.
tradingstrategies.work/dashboard/bi...
Study the Past — Improve your Future 🥋 #backtesting
@tradingstrategies.work9/9
Four questions for any RWA number:
→ in the vehicle, or on the chain?
→ distributed, or represented?
→ same tracker, same boundaries?
→ which date?
tradingstrategies.work/dashboard/cr...
Not investment advice, not a forecast.
Study the Past — Improve your Future 🥋
#backtesting
@beefedai.bsky.socialMonitoring and Continuous Optimization of Investment Algorithms
https://beefed.ai/en/monitor-optimize-investment-algorithms
#AlgorithmMonitoring #ModelDriftDetection #Backtesting #AbTesting #PerformanceMetrics
@tradingstrategies.work9/9
Before believing any "after X came Y": ask for the drift, the pairing, the independent count, the locked definition.
Run any state:
tradingstrategies.work/market/bitco...
Not investment advice, not a forecast.
Study the Past — Improve your Future 🥋
#backtesting
@tradingstrategies.work1/9
Bitcoin made a median +31 % in the six months after quiet volatility.
It made +27 % in the same six months without the condition.
The first number gets posted. The second one decides whether the first means anything.
@tradingstrategies.work10/10
Watch not how many allocators agree, but what share of Bitcoin sits where nobody made a Bitcoin decision.
tradingstrategies.work/blog/bitcoin...
Not investment advice, not a forecast.
Study the Past — Improve your Future 🥋
#backtesting
@tradingstrategies.work10/10
A flatter ride, in every case measured. Paid for in the rally windows.
Open: what sets how much rally you keep?
tradingstrategies.work/blog/grid-bo...
Not investment advice, not a forecast.
Study the Past — Improve your Future 🥋
#backtesting
@tradingstrategies.work1/10
Your grid bot is not a hedge. It is half your money sitting in cash.
That is where the protection comes from, in every one of 138 windows measured since 2017. And it is why the bot misses the rally when it comes.
