The real difference between results vs PRT

Forums › ProRealTime English forum › ProOrder support › The real difference between results vs PRT

Viewing 8 posts - 1 through 8 (of 8 total)
  • 04/23/2020 at 11:20 PM #127699
    jebus89

    So… i exported all my trades from 01.01.2019 -> 23.04.2020 from IG and placed it in a pivot table to view some results and compare it with the Detailed report stats in PRT and the findings are.. pretty shocking. 

    Im exporting the IG.com trade-history list called (translated) “Profit/Loss overview” and i exported the data into excel and im viewing the column called P/L (and not the one called Total which calculates your total P/L after cost is subtracted.)

    What you see in the screenshot is: The difference between IG P/L on each market – PRT P/L on each market, and as you can see the difference is total -574€

    Meaning that only looking at PRT detailed report, you would think I was 574€ richer than i am. Not only that but the number of trades (marked in red) where off by +\-1. This could be some sort of accident/bug or maybe related to when a trade was opened and closed compared to the dates i used for extracting the data, but 1 trade should not be effecting the stats overall so much so its pretty safe to say that the difference is significant without the +\- 1 trade.

    Lets look at Wall st market as an example: From 407 trades, the difference between “IG numbers – PRT numbers” = -266 €. I trade 1€ contracts so this is a total of 266 PIPS!! difference… This is from multiple strategies.

    When comparing numbers WITH all the trading related cost included, (meaning im looking at the column thats called “Total” from the IG report) the difference between IG – PRT = -377 €

    Now, i understand (kind of annoying but..) that PRT cannot calculate all costs related to trading (like holding overnight etc), but thats “only” -111 € difference between “Total P/L” VS “P/L”.

    Meaning still that the missing -266€ is just purely from PRT “guesstimating” your actual buy/sell price vs IG’s actual buy/sell numbers.

    So what does all this mean? Well when your looking at a backtest, just know that result is actually gonna be “alot” (depending on who you ask i guess) worse. The backtest cannot be 100% trusted. Im not rich, and my trading account would have LOVED the missing -574€ total missing…

     

    Please, check these numbers for yourself and share them here if your findings are similar to me so we can compare. Also if you think ive calculated something wrong or have questions on how i did it, let me know and ill answer you.

    Step 1. Go into IG history – Download the P/L Overview list from a selected date of choice (preferebly as far back as your detailed PRT report allows you to go)

    Step 2. Convert the data to excel and create a Pivot table (let me know if you need help doing this) – I needed to tweak the , vs . in excel to get the numbers right.

    Step 3. In the pivot table, use “Market” in Row and Sum of P/L + Count of P/L as the Values (meaning the Sums P/L of each market + amount of trades on each market)

    Step 4. Compare your IG report for each market with the detailed report from IG from each of the same markets, using the same dates as your IG report is from.

     

     

    Edit: This also means if youre running multiple strategies on 1 market, IG cannot show you the true results for each strategy because it can only look at the Market as a whole and not each strategy alone. PRT can look at each strategy alone, but the numbers cannot be trusted 100%..

    1 user thanked author for this post.
    avatar bertrandpinoy
    04/23/2020 at 11:58 PM #127701
    Vonasi

    So the gist of it is that we can’t put much faith in back testing during volatile times because we have to use a fixed spread and spread is far from fixed. We can’t put any faith in PRT forward test results either because they don’t know what actual trade prices we got. So the only thing we can put faith in is the bottom line on our IG demo accounts or real accounts but we cannot know from them exactly how much each strategy added to that bottom line and which took away from it unless we only have one strategy on each instrument. I guess that means that if we want multiple strategies on an instrument then we should not forward test or go live with strategies that are not showing very high gains for each trade….. damn IG for not passing the info we truly need back to PRT…. anyone would think they wanted us to lose! 🙂

    04/24/2020 at 12:23 AM #127702
    jebus89

    Yes @Vonasi, pretty much you need to run strategies with “big” wins on each trade (avg trade) and not something with a “small edge” for each trade.

    I dont have enough trades to look at during non-volatile times, i started trading with IG/PRT for real in october 2018 when volatility was already pretty high and just before the big -20% drop in 2018..

    If someone with trade data from before 2018 could do the same test, that would be great, so we can maybe get a more clear view of the difference when volatility is less then it is now.

    The issue regarding not knowing what strategy is performing well if you run multiple strategies on the same market. I see that you can open multiple “accounts” for live trading within your IG account, i wonder if its possible to run 1 strategy on 1 market per account? Or if it just opens the same PRT account when you open PRT on your “new cfd account”.

    Im talking about on “My IG” page it says “Add account – Trade barriers, CFD, mt4”. Gonna send an email to IG tomorrow and ask if this is possible. Which would obviously suck because you need to allocate money into each of the accounts, so i guess its not really an option if you got like 10 systems on 1 market… You would need to fill up each account with enough money to last through a “worst drawdown period” + some for the margin on the next trade after that….

     

    Edit: i see a typing error on Step 4. I ment to say Compare you IG report to your PRT detailed report.

    04/24/2020 at 8:16 AM #127708
    jebus89

    I wonder if switching from market order to limit order would fix most of this?

    Lets theoreticly say instead of market orders (slippage) you use limit orders and you get filled on every trade as you would with market orders. this would remove slippage as you are guaranteed fill only on that spesific time..

     

    anyone using Limit orders only?

    04/24/2020 at 9:58 AM #127739
    Nicolas

    Backtest of the ProRealTime software does not include overnight or overweek fees from your broker. You can either simulate it by adding fees into the backtest settings, thought that you knew that?

    04/24/2020 at 10:24 AM #127744
    jebus89
    Nicolas wrote:

    Backtest of the ProRealTime software does not include overnight or overweek fees from your broker. You can either simulate it by adding fees into the backtest settings, thought that you knew that?

    Nicolas: As i said in my first post, the cost of trading is just -111€ (for my wall st algos)

    My quote from first post: “Now, i understand (kind of annoying but..) that PRT cannot calculate all costs related to trading (like holding overnight etc), but thats “only” -111 € difference between IG “Total P/L” VS “P/L”.”

    The problem isnt the cost, the problem is the pure P/L of a trade, without cost included. The difference comes from PRT guesstimating your buy and sell price vs the price you are actually getting (buy/sell).

    So for example lets say you do 1 trade, you hold it only for 1 hour, so no overnight cost etc. PRT tells you that you just got 100 pips profit from your last trade, but then you check IG and IG says you only got 98 pips of profit. Add all of those “lost pips” and you get a bunch of money that you thought you had, by looking at the PRT report, but if you start looking at the IG numbers, the real numbers, it is in fact a lot of money lost.

    04/26/2020 at 7:50 PM #128268
    TempusFugit

    Hi Jebus,

    I use limit orders with several of my systems. I have analyzed the differences between market and limit orders quite a lot, but not with backtesting as you can´t trust their results about that. For that I have run a  lot of systems in Demo with two versions market and limit orders being the only difference to compare results. My conclusions are that the limit orders tend to be better but not always. With limit orders in most of the trades you are going to get a little better result but you are going to miss several trades that tend to be good ones.

    IMO the best type of order depends also of the system. The bigger the Timeframe the better for limit orders as there are more chances that the order is filled, And also the more trades the system makes the better for limit orders because if the system does very few trades you don´t want to miss any that can be the key for a good profit.

    For reference if you are in a 1h Timeframe with a not big spread product (DAX, DOW, EURUSD…) using “AT CLOSE LIMIT” orders will make you miss about 5% of trades

    I usually use in my systems this snippet than allow me to backtest/define more precisily my entry:

    With “NENTRYLIMIT ” I define how many bars after the entrysignal the trade can be made at the “close limit” of the entrybar, so with more bars to fullfill the order there are more chances to get it.

    With “MARKETENTRY=1” if the order is not fullfiled after the bars defined in NENTRYLIMIT  then a market order is executed, with “MARKETENTRY=0” there is no market order

     

    4 users thanked author for this post.
    avatar Edmond, avatar jebus89, avatar Nicolas, avatar Paul
    05/26/2021 at 8:32 AM #170437
    jebus89

    @TempusFugit Thanks for the reply! Im sorry I didnt see your reply earlier.

    This is confirming what i thought, missing 5-10% of trades, and bigger timeframe = more chance to hit limit.

  • Author
    Posts
Viewing 8 posts - 1 through 8 (of 8 total)

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
ft100 vs fdax – futures vs cash – ig vs ib vs saxo
prt personal account vs prt ig account
little discrepancy between the code and the default prt pivot points
stochastic, prt vs pine script
time difference between ig and my computer
6 jan 24 prt not available – too long to respond
accumulated difference
backtest giving strange results
backtest shows wrong results
backtest vs live trade

Create your free account now and post your request to benefit from the help of the community
Register or Login