Algo Trading in India |

Data is the new oil, at Squareoff we firmly believe there are lot of hidden gems if we dig deep with the data. For months we have been working on this machine learning bot to generate a consistent returns with higher winning accuracy with least risk.

We did forward test with the ML bot by running it live in our own account with 2 lacs capital, 1 lot in Nifty and 1 lot in Bank Nifty weekly options, since Sep 2020, it generated around 45% returns in last two months with winning accuracy of greater than 65%. Please note that its directional option selling strategy.

How does Machine learning bot works?

Unlike our other trading bots which works based on price patterns and other historical data, ML bot works based on live data that includes everything price, volume, Open interest, volatility, we added as much as data we can, so that the BOT can quickly analyze the datasets and make buy/sell decisions. 

By trading with 2 lacs capital, the bot has given only 9% drawdown so far. The max loss happened so far in a day is 10k with Nifty & 10k with Bank Nifty and average number of trades in a day will be around 2 or 3 trades. We do not have backtest for ML bot, the structure is so complex that its hard to backtest like any other simple strategy, the bot was trained to find alpha with huge historical intraday datasets and later we forward tested it by deploying our own capital. So the give info is for short period of forward test sample days only, there are days where the ML bot could trigger multiple trades (as high as 12 to 15 trades in a day) and it never uses any targets but has predefined stop loss of 25% as a part of our risk management. we recommend users to trade with only just one or two lots for few months and gradually scale up the size. Trade this bot only with the capital you can afford to lose, we see ML bot could be a big game changer with dynamic analysis, over the longer run it should generate higher returns.


The ML bot trades on both Nifty and Bank Nifty, based on the live data it shorts ATM CE or PE, and keeps a risk management stop loss with each trade. The ML bot will keep changing the positions from CE to PE or PE to CE based on incoming data. It works like Stop and Reverse system, if the data suggest to short Call option, it shorts ATM call option and waits for PE signal to appear, if the data suggests to short Put option, then it will squareoff the existing call option, cancel the existing Call option stop loss and initiates PUT option strike. 

The default stop loss is kept as a part of risk management, in case of any issues from broker side due to which connection gets lost, then ML bot users will be at risk if we don’t keep a stop loss in the order book, thats why bot places default stop loss, but when next order is triggered, those SL orders are cancelled along with squaring off open positions. We recommend to trade this bot with minimum 2 lacs as capital per lot.


Machine learning bot will be available for all our existing bot subscribers without any additional cost. If you are not a subscriber, then you can access all our 15 trading bots at Rs.29500 per year. You can subscribe using this link You cannot subscribe only to ML bot, as its not a standalone bot. Currently, it works only with Aliceblue and Angel Broking. You can take free trial using this link

We always recommend subscribing to all bots instead of individual bots because, at Square off we always work on innovative trading ideas, so we keep developing new strategies and implement it as a bot, all bots subscribers will get access to any new bots we develop without any additional cost.


The Machine Learning bot has made around 80 lacs over all profits which is 35% returns in last 7  months (as on June 2021), however fewe months was rough, due to market volatility back to back stop loss was hit and the strategy was in drawdown phase, but eventually it recovered from drawdown and made new equity high . That’s why we always say be disciplined and stick to the trading bot for longer period, users who see streaks of losses would stop trading the bot thinking it stopped working, and users who traded with over leverage would have blown up their capital during drawdown phase. But all bot users who traded with correct leverage was able to sail through drawdown phase easily. We recommend to trade with 4 lacs per lot to trade this strategy. live profits of ML bot can be checked here

Profits made for one lot

Month wise PnL for one lot


29 comments on “Machine Learning Trading bot

  1. Last few days combined ml bot has given -17% on normal margin. I used 4x leverage and more than half of my capital is now gone. Started with 10 l now it’s less than 4l. How can I recover my losses?

    1. manual trading is the best, atleast you have some control. machines do not bother whether you make profit or loss.

      1. You should not put a trade without understanding the risk involved. It is your mistake to increase the leverage. Do not complain about Bots or Computer. They are far superior to our monkey minds.

  2. Sir today also heavy loss. Total 4% loss is nrml margin total 25% in nrml last 7 days. 80% of my capital wiped out in last 5 days. I am taking loan now to recover

    1. Do not over leverage, recommended capital is at least 2 lacs per lot, higher the leverage, higher the risk. And completely avoid trading based on borrowed money. That’s a big blunder one can do.

  3. What is ROI per month, if i started with Rs.2 lacs capital.
    Also plz publish every bot’s performance with minimum capital requirement for each trader

  4. manual trading is the best, atleast you have some control. machines do not bother whether you make profit or loss.

    1. Machine learning bot involves high level complex concepts where tons of data is feed into the machine and bot chooses it own rule based on the data, we do not feed in any rules.

      1. I am closely observing your bots performance. Ever day a every a different bot is winner. Do you understand the traders psyche.

  5. i think it is not advisable to trade any system at least with minimum 4 to 5 years of back test data

    1. Yes, that’s why most of trading bots are backtested with vast number of data, however Machine learning bot is something completely different from tradtional strategy, which couldn’t be backtested, that’s why we recommended users to go with limited capital initially and then scale up, and we are really gald that ML bot is doign well so far

  6. Hi, two lakh per lot would be for options right? For eg, for nifty one lot option would be around say Rs. 40000, remaining amount ie Rs. 1.60 lakh would sit idle?
    And the returns shown on telegram in percentage terms would be on Rs. 40k right? Kindly clarify.

  7. Monthly data seems incorrect. Given the large DD, ML bot should be making losses in March or April month.

    1. Drawdown was recovered by the month end, pnl is calculated from month start to end, and drawdown occured during the month recovers.

  8. I didn’t understand why performance is showing in heavy numbers in telegram. Show it how it will be per lot basis.

  9. Hi. I am part of ur telegram group for many months. I was considering to sign up but I have observed a major issue where, if I consider the daily profit % shared in the telegram group and assume I allocated say 100/- for each of the 14 bots every day, then I see that in 5 month period ending in June 2021, we would be in loss because some bots don’t trigger hence more capital used. Since I am using same amount say 100/-or 3lakhs for each bot daily, I ca calculate profit by simply adding all positive daily % for all bots & deduct all negative % of that day from it. So I take that return% and keep adding such calculated figure daily for last several months (ending June 2021), it shows losses & also less return since many bots don’t use capital everyday.
    Infact this is more scientific way to report results than how much each bot made every day because this number is a function of how much capital was allocated by users on that day.

    So Going by my calculation, I am confused how it’s possible that returns are in negative??

    May be profit is function of which bot people are putting money most times. Or something else must be missing. But I calculated based On daily reprot shared in Telegram and it showed negative return. Very confused. Can you do the calculation I mentioned above to show net nav daily if some allocates 100/- daily on each of the bots? This would be scientific way to do. Because people are not that much interested in how much all users made but on how much return all bots generated over time of equal amount was allocated to them daily. For this calculation, you can exclude bots which are designed to work on specific days of the week.

    Please share this as I am considering to deploy my capital into this.

  10. Please create one Nifty Straddle Bot with 25% and 40% SL @ 10.30, by 10.30 all noise gone and market settle to move one direction, Drawdown is very less compared to 9.20 straddle..

Leave a Reply

Your email address will not be published.