Two bots were run on identical settings a dozen seconds apart. Both operate in muliti mode, simultaneously tracking about 25 cryptocurrencies and executing transactions when conditions are met.
Problem:
The transactions executed are largely different. Bots operating simultaneously often choose different currency pairs and entry/exit moments.
Below is a snapshot. In green highlighted transactions that repeated. Should repeat all, or at least most of them.
The results of the transactions, once they occur, are similar.
Snapshot: boty_porownanie.jpg (1693×855)
Test bot: Gainium app
Real bot: Gainium app
Two small corrections
1 Test bot and Real bot headers are swapped. Sorry 
2 I found one more correct transaction unchecked, Shrap at 01.45
However, it does not change the situation
Multi coins bots paper and real are not always the same. You have TP and SL based on %, so exit signals depends on entry price and fee, which is slightly different in real and paper. As I can see they are started pretty much similar, but later entry price has difference (maybe also fee, didn’t check)
Thanks for the quick reply
- SL does not matter in this case. It has 10% and only protects against absolute disaster.
- OK, I understand that there may be different output signals. But why is there a different INPUT?
- do you have any suggestions on how these results can be approximated?
This is also case if you will create 2 real or 2 paper bots. The deals inside might have different entry prices
Input are the same, they are using the same indicators data.The reason why one bot started and other doens’t is in max deals limit. In paper, for example, there is a free slot on SPARTA, but in real bot the deal on SPARTA is still active, so it wont open new one.
The best way to make results stable is to exit by indicators, in this case entry and exit signals always will be the same.
1 Like
As I understand it then:
The differences between the paper market and the real market cause the old transaction not to be completed, which blocks the opening of a new transaction.
So we have two ways to fix the situation.
1 Faster - increase the limit of possible transactions
2 Better - change the signals from price movement to indicators
If I make a mistake in reasoning please correct me.
And thank you very much for your quick response. To start with, I will try the faster way and report on the results.
1 Like
Below is a new comparison.
In the previous version, the bot allowed 3, now 6 simultaneous transactions
According to backtests, this is enough to miss less than 1% of possible deals
The effect, unfortunately, unsatisfactory
- the list of bot transactions still differs. In both bots there are transactions that have no equivalent in the other one
- exit times are also different, but this I can understand
The bots have been stopped in order not to generate losses
Screenshot: https://uploadfile.pl/pobierz/2325103---f0pm/Test-real_2.jpg
Bot real: Gainium app
Bot paper: Gainium app
