Reliability of streak algo in real-time markets?


#1

I created an algo in streak for Intraday, which gave 9.3 % return after accounting for brokerage, data was backtested for 3 months, as is possible on a 15 min intraday candle.

But when i deploy it, i am making losses, so, i wonder what is the reliability of such algos?

@Streak


#2

What was the drawdown of the strategy and Win to loss Ratio?

How many days has it been since you have deployed the algo ? are you taking all the trades that come along ?


#3

Drawdown = -3.33%
Win : loss ratio = 71: 67

Today is the first day, first two trades results in a loss.
Yes, i took the trade when it alerted me to, both the times.


#4

Bro since your drawdown is -3.33, this means that largest dropdown in your portfolio during backtest was 3.33% which should also be expected when it is deployed in live markets. For risk management purposes, professionals generally expect double the drawdown in their portfolio to account for worst case scenario.

Also, 9% profit was generated over 3 months, you have used the strategy only for one day. You need to give some time to check if the strategy is really working or not.Your win/loss is 71:67, which is almost 50% that means out of all the trades you take, only 50% will be successful. You might just be having a bad day today.

You also need to understand that strategies do not perform similarly during all market conditions hence it is necessary to run the backtest in different periods and validate it.


#5

Thanks for replying,

Could you let me know the ideal win:loss ratio?
Also, if the strategy fails with other scrips, should i dump it?


#6

Iā€™m not a professional adviser but ill share whatever I know through my experience.

There is no ideal win/loss ratio as such. You can be a non-successful trader even if your win/loss ratio is 90% and a successful one even with a 10% win/loss ratio. Let me explain how. Suppose your average win per winning trade is Rs. 1 and your average loss per loosing trade is Rs. 10, now if you take 10 trades, according to win/loss probability, 9 should be successful and you make rupees 9. But the moment 1 trade goes wrong, you will loose all the profit from 9 trades and make a loss of Rs.1. Compare this with other trader who has only 10% accuracy but takes 10 rupees profit and 1 rupees loss in winning and loosing trades respectively.

So it is necessary to compare your Win/Loss ratio with Average Gain per winning trade, Average loss per loosing trade and your Risk/reward which is basically your take profit and stoploss. The same theory applies to Risk/Reward which I explained above.

Hope this was helpful and share it if possible so that more and more people can learn.


#7

Regarding your other question, if the strategy fails with other scrips, should i dump it?

Different stocks have a different movement. Because they are affected by different fundamentals and different type of traders. A stocks movement and patterns are created by the traders who trade them and there are different types of traders present in different markets (Eg. scalpers, Mean reversion traders, Trend traders, swing traders etc.). For various reasons different group of traders choose to trade different instruments. It is the inherent behavior of these groups that is responsible for characteristic pattern in movement of a particular instrument. Hence pattern in movement of different instruments differ and not all strategies perform similarly on different instruments.

Well ideally a good strategy is the one that is also robust. This means that returns, drawdown etc. should not change a lot. Some variation is obviously expected but it should not give completely opposite results.

When you are trading in markets, you are competing with the market and the market consists of people who participate in it and among those people are some of the best brains hence you need to act really smart to make consistent profits.


#8

Yes, this was helpful, i will share it with others i encounter.