Does Algo trading really work?

[lakshmibala]

Latest back test results for the same period by the same algorithm attached, Don't get surprised on check the returns % compared to the old one

 

[pannet1]

Algo with Zerodha does not has good accuracy on the results irrespective of P/L,
With Any algo platform, the results should be constant for same period being back test irrespective of time for the indicators which are independent of time, But it is not so, Hence I have lost the confidence with streak.tech algo,
Same algo used yesterday which gave 110% results in 90days back test period is showing -15% results now,
Same algo used in live market gave a buy in HUL at 1803 is giving buy only at 1780 in the next day when I back-tested it after a day
Same algo with same backtest period is giving different results, hence totally lost the confidence with streak.tech algo system
I used EMA breakout system

talk to a professional algo making company

1) how u trade
2) algoji

to name a few. if you tell them the strategy and the backtest result, they will tell you what are the pitfalls. you see this needs experience of automating many strategies.

edit: backtesting assumes that an order is not fired beyond the OHLC in that timeframe, but real world is very different than this. add slippage. try altering your strategy with a % based stop loss and let the winner run with a time stop.

 

[lakshmibala]

I traded 1 month with the algo which was having back test results of 55% in hind uniliver, the issue with the Algo system with streaktech is " results are not consistent for the same algo for the same period of back test" Now it shows 15% results, some time it shows -10% results
I don't mind if the algo is resulting loss or negative results, but it should be same all the time, but it is not so,
Then how can one trust with streak tech & proceed further based on the backtest results,
I am atatching a backtest results which was done on 24th December, you will get surprised to see it, But if I back test the same algo for the same period it gives different results, Check the attached screenshots
I was in excitement when i found a super successful algo to make 100% returns using the it with in 60days, Later I realized that its not at all consistent with streak tech, People like me should do lots of backtesting for same period in different occasions before coming to conclusion, and multiple different periods when they find a good algo to know the consistency in results

 

[vaibhav14476]

I have a short experience with Algo trading but not a very pleasant one. I read recently that operators know when algos will be triggered to give directional calls and when stop losses will get triggered. So they will first drive up the prices and then quickly move the prices down to trigger the algo on the wrong side. Hence it's quite often that stop losses get triggered soon after which the stocks move back in the right direction again.
Is that something the others have also experienced?

First of all, algo should be used with big volume counters like nifty, bnf, Mac two crude etc. I myself back tested ma cross strategy on each & every trade on these 3 counters with hourly, 4 hour & daily timeframe for last 3 years. It works. Now I m doing profitable Trading in crude on 4 hour timeframe. No stop loss. Just converting position from buy to sell then again buy & so on. No of lossmaking trades are more than profit making trades. But losses are small and profits are big. Important is to take each and every trade.

 

[creativegene]

Algotrading works only if you build your algorithm and backtest on historic data yourself or someone truly does on your behalf. The general system of the market do not have a true backtest done where they don't consider fill rate or slippage etc.
But all said yes the system works but it's a lot of work to build a system that do perform and then testing it with real money is something you have to do for surely knowing it works.