Backtest vs Papertrading Results from Streak - Pretty Close

Have been using Streak for couple of months now. After designing and fine tuning the strategies based on backtests, started paper trading with few issues (coding as well as platform issues) uncovered, work arounds implemented etc…here are results of backtesting vs Papertrading results. This is for benefit of those who want to explore Streak and always wondered abt how close backtesting vs paper trading vs live trading results.

NOTE:

  1. Live Trading results will be further different (positive as well as negative) due to additional factors in play but this should give a brief picture…
  2. All strategies are option buying strategies. Multiple versions with all Buy and Sell strategies having similar entry and exit points but with minor difference.
  3. The strategies are designed with 10X, i,e with 10 lots of Nifty for better picture.
  4. The PNL of both Backtesting and Paper trading is Gross PNL which includes brokerage and other charges. Since paper trading doesn’t provide Net PNL, Gross PNL has been used for better comparison.
  5. The strategies are deployed with Market price based execution. Limit Price based execution is generally less preferred from what I have observed among other strategy creators as well.

Its a very small test period. But it is enough to give a picture of just how close backtesting results are with actual trading. While it does make a difference, the results seem pretty close when we consider overall period. This is pretty good when we consider the significant role the backtesting plays in designing the strategies and its definitely one of the strong points of Streak platform.

However, I would like to provide the following feedback to @Krishnendu and @Streak team.

Couple of missing features which played a major role in strategies being less successful.

  1. LTP based signalling -
  • The lack of LTP based signalling caused many opportunities to be missed (because the entry condition does not satisfy on close of the candle). While the team considers such cases as false signals, it actually missed many good opportunities in my case.

  • Also, since it is market price based execution, it led to entries at higher points than intended and as a result hit the stop loss more frequently. The alternative of using limit price based execution is not much helpful either since it misses quite a few opportunities even with buffer in place.

Telegram Streak support team has recently confirmed that some sort of LTP signaling option without waiting for Candle close. Hoping that it indeed helps many of us who have been requesting for such option.

  1. The ability to adjust positions after entry -

Since its difficult to capture the perfect entry, it becomes very important to fine tune that entry by taking multiple entries. The lack of this ability constraints the flexibility and efficiency of the strategies to a great extent.

Though it is not apple to apple comparison, for the purpose of demonstrating the effect of having the above two features built into the strategy, here are results from another platform which supports the above features. The success ratio increases quite significantly . The entry and exit points are exactly the same (Nifty Buy 01 vs Nifty Buy 01) but with LTP based signals and further adjustments done after entry.

image

Fellow Streak users…Please feel free to provide the details of your experience and feedback.

Not to forget the initial feedback :wink:

noted @Magnum007

Please elaborate on this. What type of adjustments are you referring to? only price averaging by multiple entries?

1 Like

Hi @Krishnendu ,

Yes in this particular strategy, I have used the “Adjustments” or"repair" feature of the platform to take multiple entries. i would assume its easier to implement, as it would be just another condition to be added to the strategy structure… Based on the condition reenter the same Contract and same quantity as specified in the entry.

However, keeping its future and other potential uses in mind, especially option selling strategies like straddle where strikes are adjusted based on movement, the feature can be made more flexible by giving an option to have its own Contract selection area

Or does Streak have any other direction (designwise) in mind in the roadmap?

Best regards,
Kiran

Hi @Magnum007 ,

We have some new design-wise aspects in our roadmap.
The challenges are making sure backtesting is accurate to model the live results and making sure there are no open ends. The team has set an approach for this and you can see once it is rolled out.

Regards

Cool! Thanks…Good to know that there are plans in this direction…There are lot of algo users who are just waiting for these flexibilities in streak as the alternate platform is not very reliable but sticking to it since there is no good alternative. Hope to see these features rolled out in Streak in short term future…

Hi @Krishnendu, its been almost 2 years now…Any of the limitations addressed?