ctid211782
03 Apr 2016, 14:23

RE: RE: RE: RE: RE:

Apologies for spamming the page like that, not what I intended, but I should have realised.

Moderators if you see this please remove all but the last of my posts.

Tim

ctid211782
03 Apr 2016, 14:13 ( Updated at: 23 Jan 2024, 13:16 )

RE: RE: RE: RE:

ctid211782 said:

ctid211782 said:

ctid211782 said:

 

 

 

 

 

Spotware said:

Dear Traders,

The issue has been resolved. We kindly ask you to remove your backtesting cache which is located in the following path:

C:\Users\%UserName%\AppData\Roaming\%Broker Name% cAlgo\BacktestingCache

We apologize for any inconvenience.

 

Hi,

I'm, still having problems (in April 2016) very similar to those described by [Stermy] earlier in this thread. 

It is interesting that the issue is attributed to data problems here.

I suspect a problem caused by cAlgo backtesting the results of massive amounts of ticks simultaneously (or close to it) across a high number of threads across multiple processors/cores. This seems to be leading to sentinel values not being incremented in time to be considered in the evaluation of bot logic in other threads.

My reason for this thinking is as follows:

Just like [Stermy] my backtest made and lost unfeasible amounts of money very quickly, making 100 million over a month in one case, and having massive drawdown very quickly in another.

I was testing a bot which evaluated trade opportunities in the onTick() function, and I was testing it against tick data.

What I figured out is that the bot was entering hundreds of trades where there is logic and sentinel values within the code which should limit the bot to a single trade at a time.

I tested this by printing the sentinel values to a log and confirmed that the sentinal values weren't being updated between ticks.

Because of this a hundred or more ticks in a row would sometimes meet the entry criteria and would trigger a hundred trades where the bot should only enter 1 trade. This entering of hundreds of trades where one trade was intended explained the massive gains and losses.

Maybe I've totally missed the mark with my theory, but either way, heads up that the previous problems (that are reported as resolved) doesn't seem to be resolved.

Cheers,

Tim

 

 

ctid211782
03 Apr 2016, 14:12 ( Updated at: 23 Jan 2024, 13:16 )

RE: RE: RE:

ctid211782 said:

ctid211782 said:

 

 

 

 

 

Spotware said:

Dear Traders,

The issue has been resolved. We kindly ask you to remove your backtesting cache which is located in the following path:

C:\Users\%UserName%\AppData\Roaming\%Broker Name% cAlgo\BacktestingCache

We apologize for any inconvenience.

 

Hi,

I'm, still having problems (in April 2016) very similar to those described by [Stermy] earlier in this thread. 

It is interesting that the issue is attributed to data problems here.

I suspect a problem caused by cAlgo backtesting the results of massive amounts of ticks simultaneously (or close to it) across a high number of threads across multiple processors/cores. This seems to be leading to sentinel values not being incremented in time to be considered in the evaluation of bot logic in other threads.

My reason for this thinking is as follows:

Just like Steve in the original post I made and lost unfeasible amounts of money very quickly, making 100 million over a month in one case, and having massive drawdown very quickly in another.

I was testing a bot which evaluated trade opportunities in the onTick() function, and I was testing it against tick data.

What I figured out is that the bot was entering hundreds of trades where there is logic and sentinel values within the code which should limit the bot to a single trade at a time.

I tested this by printing the sentinel values to a log and confirmed that the sentinal values weren't being updated between ticks.

Because of this a hundred or more ticks in a row would sometimes meet the entry criteria and would trigger a hundred trades where the bot should only enter 1 trade. This entering of hundreds of trades where one trade was intended explained the massive gains and losses.

Maybe I've totally missed the mark with my theory, but either way, heads up that the previous problems (that are reported as resolved) doesn't seem to be resolved.

Cheers,

Tim

 

ctid211782
03 Apr 2016, 14:12 ( Updated at: 23 Jan 2024, 13:16 )

RE: RE:

ctid211782 said:

 

 

 

 

 

Spotware said:

Dear Traders,

The issue has been resolved. We kindly ask you to remove your backtesting cache which is located in the following path:

C:\Users\%UserName%\AppData\Roaming\%Broker Name% cAlgo\BacktestingCache

We apologize for any inconvenience.

 

Hi,

I'm, still having problems (in April 2016) very similar to those described by [Stermy] in the original post in this thread. 

It is interesting that the issue is attributed to data problems here.

I suspect a problem caused by cAlgo backtesting the results of massive amounts of ticks simultaneously (or close to it) across a high number of threads across multiple processors/cores. This seems to be leading to sentinel values not being incremented in time to be considered in the evaluation of bot logic in other threads.

My reason for this thinking is as follows:

Just like Steve in the original post I made and lost unfeasible amounts of money very quickly, making 100 million over a month in one case, and having massive drawdown very quickly in another.

I was testing a bot which evaluated trade opportunities in the onTick() function, and I was testing it against tick data.

What I figured out is that the bot was entering hundreds of trades where there is logic and sentinel values within the code which should limit the bot to a single trade at a time.

I tested this by printing the sentinel values to a log and confirmed that the sentinal values weren't being updated between ticks.

Because of this a hundred or more ticks in a row would sometimes meet the entry criteria and would trigger a hundred trades where the bot should only enter 1 trade. This entering of hundreds of trades where one trade was intended explained the massive gains and losses.

Maybe I've totally missed the mark with my theory, but either way, heads up that the previous problems (that are reported as resolved) doesn't seem to be resolved.

Cheers,

Tim

ctid211782
03 Apr 2016, 14:09 ( Updated at: 23 Jan 2024, 13:16 )

RE:

Spotware said:

Dear Traders,

The issue has been resolved. We kindly ask you to remove your backtesting cache which is located in the following path:

C:\Users\%UserName%\AppData\Roaming\%Broker Name% cAlgo\BacktestingCache

We apologize for any inconvenience.

Hi,

I'm, still having problems (in April 2016) very similar to those described by [stevefromnaki] in the original post in this thread. 

It is interesting that the issue is attributed to data problems here.

I suspect a problem caused by cAlgo backtesting the results of massive amounts of ticks simultaneously (or close to it) across a high number of threads across multiple processors/cores. This seems to be leading to sentinel values not being incremented in time to be considered in the evaluation of bot logic in other threads.

My reason for this thinking is as follows:

Just like Steve in the original post I made and lost unfeasible amounts of money very quickly, making 100 million over a month in one case, and having massive drawdown very quickly in another.

I was testing a bot which evaluated trade opportunities in the onTick() function, and I was testing it against tick data.

What I figured out is that the bot was entering hundreds of trades where there is logic and sentinel values within the code which should limit the bot to a single trade at a time.

I tested this by printing the sentinel values to a log and confirmed that the sentinal values weren't being updated between ticks.

Because of this a hundred or more ticks in a row would sometimes meet the entry criteria and would trigger a hundred trades where the bot should only enter 1 trade. This entering of hundreds of trades where one trade was intended explained the massive gains and losses.

Maybe I've totally missed the mark with my theory, but either way, heads up that the previous problems (that are reported as resolved) doesn't seem to be resolved.

Cheers,

Tim

ctid211782
22 Mar 2016, 03:32

Related Idea.

Thinking along the same lines it would be great to be able to export optimisation results. (Ideally to .csv , .xml , or another excel compatible format.) for record keeping and further analysis purposes.

Cheers.