Plain words, both ways
You write "sell if it drops 5% from the high." We answer with "worst drop," not "maximum drawdown." Jargon is a tax on understanding, and you are not the one who should pay it.
About
Most trading ideas are never tested.
Only traded.
The distance between having an idea and knowing whether it ever worked is filled with Python, data plumbing and a dozen ways to fool yourself. Almost nobody crosses it. AlphaBack removes the crossing.
The problem
A week of work stands between you and a straight answer.
You hear an idea somewhere. Buy the dip. Ride the trend. Cut the losers at eight percent. It sounds reasonable — and sometimes it genuinely is.
The only honest way to find out is to run it against history. That means a data provider, pandas, a backtesting library, and a few days learning why your first result showed a 400% return. You were looking into the future. Everyone does it once.
So most people skip the step entirely. They test the idea with money instead, and they find out the same thing, slower and more expensively.
The market is a very expensive place to discover you were wrong.
Principles
Four rules we build against
Every product decision so far has come out of one of these. When they conflict with growth, they win.
The code is always there
Every run hands back the Python it actually executed. Read it, copy it, run it yourself, disagree with it. A result you cannot audit is a result you should not trust.
Real bars, real costs
Actual daily open, high, low and close. Commission taken out. Orders fill on the next bar, never the one your signal fired on. The friction is the part that decides most strategies.
Say what it can't do
A backtest is a rehearsal against one recording of the past. It is not a forecast, and a product that lets you believe otherwise has done you real harm.
The other list
Things AlphaBack will never do
Worth writing down while it is still easy to keep.
- Sell you signals. The moment a product profits from telling you what to buy, its incentives stop matching yours.
- Show a strategy's best stretch. Any idea looks brilliant over the right eleven months. You get the whole window or none of it.
- Promise a return. Not in the copy, not in the charts, not in the small print.
- Hide the mechanics. If a number is generated rather than measured, it says so on the page.
- Take money from the other side. No brokerage rebates, no order-flow arrangements, no affiliate links. Nothing that would make us prefer one answer to another.
Test the idea first.
Then decide.
One sentence in, a full report out. No card, no install, no account needed to try it.