Market order strategies and Spread

I don't necessarily know that this is a bug. Market order strategies are typically harder to find from my experience. If the spread is added in then a high amount of trades will result in the spread eating into a lot of net P/L which can tank your results and thus SQ doesn't produce strategies with more trades. I would just use a custom fitness and weight a high number of trades if you want to get more trades out of the generation process.

please attach your buid configs with and without spread, so I can reproduce it. 

As Khari Lane says, spread makes results worse, so it could be the reason.

They were build on 126 and i retested them on 128 Automatic this i forgott .Retest this gave the diffrence not the spread. This is not good either i suppose. Do you reccomend to retest all strategies and exchange all indicators in portfolios after each release ?  I mean i lost 1/3 of the strategies atleast doing this retest and 1/3 probably if i check my 31 portfolios made from 126.

it is hard to say without seeing the configs or some the strategies that passed in 126, but not in 128.

There are some bugs fixed in each build, so it is possible that your strategies worked in SQ 126 because of a bug there.

It is generally good to retest your strategies with a new build.

I agree, there are much lesser strategies with a little spread :

For exemple :

with 1 pip spread I may have 58 accepted stratégies per hour .

with 2 pip spread, I may have 8 accepted stratgies per hour.

This is why I am searching stratégies with 1 pip spread only, otherwise I will not have enough stratégies

Voted for this task.

I get very different performance in regards to evrything, number of trades, result on some types of strategies depending on settings in regards to Spread and distance to price. I do not see the need for distance to price if the strategy from beginning for example a limit order of 5 pip distance if the strategy places an order even if the distance is only 2 pip since there seems to be no rule to place it 5 pip from price in the code and having a setting when testing that has no function in reality seems unnecessary it should probably be connected to real tick test with spread instead to calculate if you got filled or not. I wondered why the limit strategies fails placing orders when i had set distance to 5 pip and still i have alot of failed limit orders placements. So the difference can be huge only by this feature both in regards to trades and result. So the best thing might be is to have 0 spread and alot of slippage and 0 distance to price and let montecarlo simulate spread and distance to price to delete the bad ones.

The results are really different for me too only with one pip !

is it because of the tic size ? When I look to the multicharts source code of one strategie , the ticksize =minmove/pricescale.

This is not correct, I had to replace it Inside the source with ticksize =0.0001. and here I could get the same results as SQ.

Maybe this is not related, but it look like the multiplier of the spread is too large :

with 1 I am getting 58 accepted strategie per hour , 

with 2 I am getting only 8 accepted stratégies per hour

with 3, I would probably have no strategie

there is too much difference

MABI is right, there is missing the MIN. DISTANCE in the MQL code, if i remove the min. distance from the strats, the backtest is accurate - https://roadmap.strategyquant.com/tasks/sq4_6555

hankeys,

As I am using multicharts engine, I don't have the min distance option and I still have this big difference

distance from price is a feature that is used by MetaTrader brokers, I'm not aware that it is used by Tradestation. 

And it is not something that should be simulated or supported in EA, it is simplky a property of a broker how close he allows the price to be to actual Bid/Ask.

Mabi - if your strategies don't match between SQ and MultiCharts then there must be some other reason for it. If you';ll create a task for it and attach a few examples we can look at it.

Voted for this task.

min. distance is the setting in the SQX itself - so the strategies should behave as we have set

for market orders the min. distance is totally nonsense, because market order is market order, so when marker building blocks are selected, the min. distance should be hidden or set to 0

for limit and stop strats - the min. distance could have logical use - if i dont want to set my pending orders too close to the price, i could use some min. distance setting

its true, that min. distance was usefull many times age, when brokers dont allow to place pending orders too close to the current price, now the limitation are over and i dont any broker which is the min. distance using for pending orders

but if we have this setting in SQX, the backtest need to be accurate and min. distance need to be solved by MQL code

Voted for this task.

When I set 0 $ slippage, I have 60 accepted stratgies per hour

When I set only 1 $ slippage, I have only 12 accepted stratgies per hour. This is 5 times less stratégies !

It is a big drop with for a 1$ slippage as well. .

Voted for this task.

slippage is in pips, not USD - and what have this setting to do with some number of accepted strats?

yes, with slippage you will get worse backtest, so by your ranking criteria could lead, that accpeted strats will be lower

Usually yes the slippage is in pips, but it can be in $ as well . (see capture attached)

It depend if you choose Metatrader engine or Tradestation/Multicharts engine.

(It would be helpful to have it in pips with Multicharts/Tradestation engine.)

I'm closing this task, there is nothing to be done. YOu'll have less accepted strategies with slippage. Slippage can be in pips or $ dependig on the engine, and it is configured the same way as in the target trading platform.

We added handling of min distance from price into the strategy code as well.