EdgekilnBacktest grading for traders
Is your backtest real, or luck?
Most great-looking backtests are tuned to the past. Edgekiln tells you, in plain English, whether your edge survives fresh data, real costs and the number of versions you tried.
Works with TradingView, MT5 and NinjaTrader exports, or any CSV of trades.
You tried 40 versions
The best of 40 random strategies looks brilliant by pure chance. Your Sharpe ratio doesn’t know how many you threw away. Ours does.
It only saw one market
A rule tuned on 2020–2021 learned that market, not an edge. We re-test it on periods and symbols it never touched.
Shuffling isn’t testing
Reshuffling your own trades can’t catch a strategy that was fitted to the past. Fresh data can.
Sample report
We graded a classic: the moving-average crossover on Bitcoin
We tried every crossover from 5/20 to 50/200 days on BTC from 2017 to 2021, 90 versions in all, and kept the best one, exactly as most traders do. Then we graded it.
BTC-USD · daily · long or flat · 0.10% fee per side
Buy when the 10-day average crosses above the 40-day; sell when it crosses back
Verdict: mostly luck
This strategy looked great because Bitcoin rose sharply from 2017 to 2021, and almost any “stay long” rule did too. On the four years it never saw, it kept only 30% of its edge and did no better than simply holding Bitcoin. And among the 90 versions tried, the winner was about as likely as not to land in the bottom half on new data. It survives higher costs, so it isn’t fragile; it just isn’t an edge.
Educational example. Data: Coinbase BTC-USD daily candles. Not investment advice. Read the full write-up, method and code.
How it works
- Bring your strategy. Upload a trade list or platform export, paste your rules or Pine Script, or build them without code.
- We test it the hard way. Unseen years and symbols, a correction for every version you tried, higher costs and wider spreads, and thousands of Monte Carlo runs.
- Get a grade in plain English. A to F, what held up, what didn’t, and why. Share the link or keep it private.
Free while we’re in early access
Grade my backtest
Send us your backtest and we’ll email you a full graded report, free. Most reports go out within 2 business days.
- TradingView “List of trades” export, MT5 or NinjaTrader report, or any CSV of trades
- Any market and timeframe
- Your file is private and deleted after 30 days
Pricing
Launching January 2027. Join the waitlist, or lock in founding-member pricing now.
Free
$0
- 1 full report a month
- Shareable link
Trader
$19/mo
- 30 reports a month
- Uploads, rule builder, Pine Script
- Re-tests on unseen data
Pro
$49/mo
- Unlimited reports
- Multi-strategy portfolios
- PDF export
Founding member
Pro for a year
$99 first year
- Everything in Pro for 12 months
- Shape what we build
- Fully refundable until launch
At launch, re-tests on fresh data cover crypto. Futures and stock re-tests are coming soon; uploaded results from any market are graded from day one.
Join the waitlist
Two quick questions help us build the right thing.
Questions
Is this investment advice?
No. Edgekiln is an educational tool that measures how a strategy behaved on historical data. It never tells you what to buy or sell, and past results don’t guarantee future ones.
Who sees my strategy?
Only us, to grade it. We never share or sell it, and uploaded files are deleted after 30 days. Reports are private unless you choose to share the link.
Why does the number of versions I tried matter?
If you test enough variations, one of them will look great by chance. We correct your result for that, the same way scientists correct for running many experiments.
How is this different from Monte Carlo tools?
Monte Carlo reshuffles the trades you already have, so it can’t tell whether those trades were fitted to the past. We also re-test your rules on data they never saw, which can.
What if I become a founding member and change my mind?
Email us before launch for a full refund, no questions asked. If we don’t launch by March 31, 2027, we refund everyone automatically.