Independent strategy research
Do grid and DCA bots actually work? We tested 70+. Every one failed.
Short version: no. We ran 70-plus grid, DCA and martingale bots through the same cost model we use on everything, on 13 years of real data. Zero survived. Grid/DCA is the only whole category in our audit that dies at a clean 100%. The equity curve these things draw is beautiful. That is the problem. It is the last thing you see before the account is gone.
The pitch, and why it is a trick
The sales copy never changes. Passive income. Works in any market. A 90-percent-plus win rate screenshotted from a demo account. All of that can be true and still worthless, because the win rate is the trick, not the edge.
Here is the mechanic. A grid ladders orders around a price and books the oscillation. A DCA or martingale bot adds to a losing position at set intervals, drags the average entry closer, and waits for a bounce. Both win small and often. Both lose rarely and enormously. The rare loss is the whole strategy. Everything else is the setup for it.
What actually happened when we tested them
We ran 70-plus of these across CME futures, tick-level FX with real bid/ask, and crypto spot and perps. Every configuration failed once real spreads and slippage were in the model. And not by a hair. Most had no edge before a cent of cost — they do not predict anything. They bet the market will not run far enough to bankrupt the position before it reverts. Over 13 years of data, it eventually always does.
The high win rate held up in the numbers. It just did not matter. A martingale closes a long stream of tiny green trades while the real loss sits in open drawdown, growing, invisible on the trade log until the day it isn’t. One move against a doubled-down position erases a year of “income” in an afternoon. We watched it happen the same way on grid after grid.
All of them? Even with wider grids?
Yes. Across futures, FX and crypto, and across grid spacing and stop settings. This is the part that matters: the failure is structural, not a tuning problem. Widening the grid buys time and makes the eventual loss bigger. Tightening it bleeds faster on cost. There is no spacing that turns a coin-flip with fat tails into an edge. On sub-15-minute FX it is worse — the spread eats the edge before it can exist, so we stopped testing FX and CFDs below 30-minute entirely.
We tried to build one that survives
We did not want to reject a whole category on principle, so we tried to save it. Hard stops. Position caps. Only genuinely mean-reverting instruments, sized small. You can keep a tight grid alive that way, as one risk-capped piece inside a larger book. But the moment you bolt on the controls that stop it blowing up, the smooth curve that sold it disappears. What is left is a mediocre mean-reversion component wearing a marketing name. That is the honest exception. It is not the passive-income bot anyone is selling you.
How we test
Every strategy is ported to Python and run against real costs — spreads and commissions modeled from tick data, not a guessed flat number. Futures come from Databento (13 years of CME), FX from Dukascopy with real bid/ask, stocks with liquidity-aware fills, crypto as spot and perps. A fast model does the bulk porting; the strongest model then tries to break every apparent winner, hunting look-ahead bias and impossible fills. We hash the code, so a strategy re-published under three names gets tested once, not three times. It is the same process that rejects roughly 78% of the strategies we test, and 100% of the grid/DCA ones.
Research and education, not financial advice. No signals, no return promises. Independent, and not affiliated with TradingView.
Which bots, and who published them?
You now know the category is a trap. What this page does not give you is the names: which specific published grid/DCA scripts we tested, who wrote them, and the exact after-cost numbers behind every rejection. That is The No List — every strategy we audited, named, with its verdict and the reason it lived or died.
Get The No List →FAQ
Do grid trading bots really work?
Not as standalone systems, in our testing. Every grid, DCA and martingale bot we audited failed a real cost model — 100% of the 70-plus we tested. The high win rate is real and irrelevant; the rare loss is the strategy.
Why do grid/DCA bots win almost every trade but still blow up?
They book many tiny winners while the loss accumulates as open drawdown, hidden on the trade log. One trend against a doubled-down position wipes out months of small gains in a single move.
Can wider grids or better settings fix it?
No. The failure is structural, not a tuning problem. Wider grids delay the loss and make it larger; tighter grids bleed faster on cost. No spacing turns a fat-tailed coin flip into an edge.
Is there any version of a grid that survives?
Only with hard stops, position caps and genuinely mean-reverting instruments — and then it is a small risk-capped component, not a passive-income bot. The controls that keep it alive remove the smooth curve that made it attractive.