ProRealTImeAlgos VS Mother of Dragons

• 08/06/2020 at 11:26 AM #140992
  StingRe

  Hi All,

  I wanted to share with you the analysis I’ve been doing on the PRTA product (paid 590 euro) vs our very own community built Mother of Dragons (big thanks to all who have contributed to this).

  Firstly, this post is not meant to represent advice or an opinion one way or another, I am merely trying to present facts and allow others to draw their own conclusions.

  I wanted to start things off by comparing two DAX algo’s. I have tried to make it as close to “like for like” as possible.

  Test Plan.

  I’m using PRTA DAX 2M Algo version 30 July 2020 Vs Mother of Dragons DAX 5M 2.2a.

  I am backtesting them on 100,000 bars only as I don’t have access to 200,000.

  I tested the same date range for both which is limited by the 2M PRTA to 22 January 2020 – 5 August 2020.

  I tested the same starting capital of 10,000 AUD with 1 pip spread.

  I tested 10 x 1A$ contracts for both.

  Attached are the results and my summary thoughts which I open for feedback. 

  1. It’s important to assess the results on a weekly basis, too many algo providers are supplying monthly views and I find it a little misleading. The reason behind this is that if you are having multiple weeks of big losses in just a couple of months it will significantly impact your trust and phycology when keeping it running.
  2. It’s important to look beyond “win rate” – win rate is deceiving and it is not a good indicator of a successful algo because your winners might be small and your looser might be big.
  3. Focus on Gain/Loss ratio – you don’t want to see the algo taking massive losses and wiping out all the gains.
  4. Understand the drawdown – again, you need to consider how much equity you are going to need to have spare to suffer losses, both in current trades and closed trades. You don’t want to see big drawdown.
  5. Time in market – the less time the better, the more time the more risk of some world event impacting your trade.
  6. Average executed orders – on the lower timeframes 1-2 trades a day is about right over a 6 month period. You don’t want too many or two few (in my opinion)
  7. Consecutive Losses – pretty obvious but you don’t want to see many consecutive losses.
  8. You want to look at what the market did in the back test timeframe and how the algos handled it. e.g Did the market go up and was it successful at taking longs vs shorts.

   

  Results (winner in bold)

  Gain: MOD – $32,700 (327%) vs PRTA – $19,193 (191.93%)
  % Of Winning Trades: PRTA – 91.46% vs (MOD -89.58%)
  Gain /Loss Ratio: MOD 3.45 vs (PRTA 2.32)
  Time in market: MOD 12.62% vs (PRTA 19.15%)
  Max Run Up: MOD $33,529 vs (PRTA $21,067)
  Max DrawDown: MOD $6,076 vs (PRTA $7,078)

  In summary, based on this test data MOD is the winner – especially if you look at the long vs shorts trades and what the market was doing when they each made their losses and gains. It looks like PRTA has issues with short trades while MOD is much better in this area and can better handle all the different types of market conditions. With negative market moves possible on the horizon I would want to be very confident that PRTA could handle them.

  I do want to thank Marco and the PRTA team for providing me with the algos for back and live testing. I think they have some great ideas but I just want people to really understand what they are buying / renting before handing over the money.

  Some other factors to consider are that we had some pretty annoying setup issues with PRTA and Timezones. They have given us a workaround for that by making a global time zone setting change but it’s not ideal. If you get this wrong, you are likely to enter trades out of hours which can add to risk.

  Hope you find this helpful.

  Cheers,

  2 users thanked author for this post.

  PLermite, GraHal

  

  StingRe

  Participant

  Topics: 3

  Replies: 20

  Been thanked: 9 times
  08/06/2020 at 11:29 AM #140997
  StingRe

  Missing attachments here. MOD closed trades stats and side by side trade comparison for all trades.

  Big thanks to @scoot3r83 for putting together the spreadsheet.

  Cheers.

  1 user thanked author for this post.

  PLermite

  

  StingRe

  Participant

  Topics: 3

  Replies: 20

  Been thanked: 9 times
  08/07/2020 at 8:02 AM #141035
  Dow Jones

  Probably another assessment I always interest to focus is the average loss vs worst loss. If distance too close, it may mean when bad trade happens, it is mostly stopped by SL. Otherwise, strategy with bigger SL and smaller TP is likely having high win rate, but hard to swallow when you get hit. When I worked on MoD DAX v2.2a, I had in mind to focus on limiting the risk and not improving the gain or win rate. I tends to limit a bit of win rate to limit weekly loss, but naturally the gain also improved. (as disclaimer, I’m not expert, but thanks to MoD, I learned a lot from it and changed some of my perspective on algo trading). I agree better to review weekly wise even though not easy to come up with strategy that satisfying weekly wise, and I also look at what is the max loss for certain week, are weekly max loss consistent (and affordable), and how many weeks are lost for the back test period. Nevertheless, I think even with MoD Dax, it is necessary that people diversify it with another strategy or market, in my opinion, the SL for MoD Dax is still quite huge and performance should be monitored closely.

  

  Dow Jones

  Participant

  Topics: 10

  Replies: 135

  Been thanked: 75 times
  08/07/2020 at 8:53 AM #141039
  Dow Jones

  Btw, when I looked at both algo weekly result chart, both algo doesn’t relate too much, if you would have run both algo, then the past 6 months, almost all the weeks are winners. It would have been a good consistent passive income 🙂

  Another perspective to look at is Mar is the distinct difference for both strategy, it may mean both of them doesn’t filter the extreme volatility during that month (MoD doesn’t for sure), so both are trading during Mar crash, MoD got lucky ride on the winners (but maybe thanks to its multiple time frame trend filtering) and PRTA unfortunately ride on losers. When both algo are high TP and high SL, it can be easily gain a lot or loss a lot. But such scenario may not repeat itself, another area to consider (although for me, I would add filter to avoid just in case similar crash happen again).

  

  Dow Jones

  Participant

  Topics: 10

  Replies: 135

  Been thanked: 75 times
  08/10/2020 at 11:56 PM #141347
  ProRealAlgos

  Thank you for this analysis StingRe.
  MoD is definitely interesting code, atleast looking at backtest, no doubt about that.

  I’ve actually had a couple of the laters versions of MoD live on my real account since april, but closed 3 out of 5 down because they didn’t perform close to their backtests results in live. In total I’m in a loss from running different versions of MoD. I believe that is because most versions of the MoD algo have been optimized using too small OOS-period. They look good in backtest but don’t perform live.

  When we optimize our algo we use a maximum 70% IS period, and often less than that. That’s why we have so good results live and not only in backtest.
  In the attached image you see the live results from MoD. If you go to prorealtimealgos.com you can see the results for the PRTA algos during the same period.

  Aside from the losses I have experienced running the algos live, I believe there is potential in MoD. Maybe in a couple of months we can see that there is a version that have performed well over a couple of months live as ours have. Until then, run the MoD algos only in demo.

  Best regards / Marco

   

  1 user thanked author for this post.

  O-jay8

  

  ProRealAlgos

  Participant

  Topics: 6

  Replies: 74

  Been thanked: 28 times
• Author
  Posts