Frosty's auto-trading bot goes live with REAL money
- Thread starter frostengine
- Start date
Hahahahaha, LOL, aw now Basis, there's no basis for that statement!
If you think an automated directional trading system which is coded such that it:
a) doesn't get aboard obvious uptrends and
b) doesn't take obvious downtrends and
c) overtrades the chop ...
in the hopes of finding a decent reversal is actually going to make you money, well, put your money on it.
It was said previously by SteveH and austinp that the bots trading needs to be based on sound trading principles. In this I agree with them, like 100%
!Not for nothing, it was very telling that when I told Frost the bot should never have initiated a Long Position during a market sell-off he actually didn't know what I was talking about, but SteveH (to separate him from any other Steve's on the Board) and austinp did, and they responded accordingly.
***
Like I said before, I'm just trying to help the kid find his way. In the meantime, what I pick-up about bots (I never would have found that info if not for looking it up as a result of this thread, BTW) may or maynot help me with developing an automated system with which I can apply any edge I might have in the markets.
***
LOL - see austinp, now I know what it's like trying to help some guy and getting flamed in response - doesn't hurt one bit, actually, it's hysterical.
Best Regards,
Jimmy Jam
Quote from frostengine:
tomorrow I will be moving to method #3........ I am probably going to end up kicking myself ... but this strategy seems more solid AT THE MOMENT so I am switching to it.....
AT THE MOMENT???
I don't understand what you are doing. Again, I am new to this, but it seems to me like you are in such a hurry to make money, that you are jumping from one system to which ever happends to be better that day.
Have you had 3 systems running this whole time, or did you tweek your system that is not working in order to try to get somthing going.
A test is a test. So far you have not tested anything because you are not leaving anything in the market long enough to know if it works.
Sorry to soud so negative, I really want this to work out for you, but I do not understand your basis for all the changes in what was supposed to be a Live test of your new system.
Edward
I have always been a very impatient person, and I am showing it here..... I'm not sure what my problem is.. I really should just stick to one strategy or pull the bot out of live action until I am ready again... Its just I know this recent performance is a bad streak... I just cant keep watching it bleed money so I try to tweak things....
its ok to be switching from one system to another. the time in the market is irrelevant. if your backtesting algorithm executes the same as paper trading (or worse estimations) then algorithms which perform significantly better than others should be the ones which are tested on the live account.
there is no need to run tests on the paper account if you've shown your backtesting is adequate. i vaguely remember an old saying about putting your best foot forward. there is no reason to test a system when you have something better waiting to be tested.
the only reason to pull the bot is if you think the extra time off the market will lead you to make significant improvements to the system. if thats not entirely expected, live testing is a bullet every system has to bite at one point.
there is no need to run tests on the paper account if you've shown your backtesting is adequate. i vaguely remember an old saying about putting your best foot forward. there is no reason to test a system when you have something better waiting to be tested.
the only reason to pull the bot is if you think the extra time off the market will lead you to make significant improvements to the system. if thats not entirely expected, live testing is a bullet every system has to bite at one point.
Considering you seem to be new at ATS development I suggest you let someone else take a look at your trade data. I think you should take advantage of offers of support from mre seasoned developers. Notwithstanding what others on this thread have written you won't be giving anything up by providing a list of entries, exits and size. I do it all of the time with a competitor of mine who is also a friend. Perhaps a different slant on the analysis will illuminate something you are missing.
Lost -$104 today.... today was sort of interesting because on 2 of the exits I closed it out early.... I am curious to see the backtest in a little while of how it would have done on its own... again WHY on earth did I even touch the trades? No clue except for a patience issue.. actually I had some justification for it.... I have been getting a LOT of screentime replaying back the days through the system and a lot of time I am able to pick better exits than the system does......
eep. another negative day? danm. why dont you set up a socket so your ats can send your trades so mine can take the opposite side. at least one of us will be making money 
jk.
i always thought it was interesting to try to make a system that consistently lost money on a no commission / spread scenario. and then just reverse the orders. makes you think about prediction in a counter intuitive way.
jk. i always thought it was interesting to try to make a system that consistently lost money on a no commission / spread scenario. and then just reverse the orders. makes you think about prediction in a counter intuitive way.
Maybe its time for a little more discussion on HOW my trades are entered.. I will not give printouts of my exact trades, but will tell a little more about the strategy
Thus far it has been classified as a counter trend strategy..... I let this classification go as it is "sort'" of close. At least when you look at the positions it enters... The strategy is more of a... umm.. I guess lets call it a reversion to bigger trend strategy... Or I guess you can call it trading the pullbacks.....
The bot trades on a VERY small time interval... what it looks for is that on a slightly bigger time interval is an underlining trend. it wants this trend to be as strong as possible... this bigger trend is not on a much bigger time interval. only slightly bigger.... which leads to the possibility that my bigger time frame trend is AGAINST what the whole day's trend is like...
So now that a trend is established there, it looks for a trend going the opposite direction on the smaller time frame.... As long as the bigger time frame's trend is still intact we will enter a position which goes with the bigger time frame or in other words against what the short time frame is doing..... This helps a lot with slippage and actually normally gives me positive slippage... because the market is still aggressively moving against where I entered....
One drawback is normally my position is in the red at least for the first couple of bars... but it then generally starts doing as it should and reverting back to the medium trend....
We only exit when the opposite signal is given or if the market appears to be becoming choppy.... I generally use ADX here.. if the adx drops below <15 I exit the position.. this is my only exit besides a reversal signal...
one of the reasons I think my manual exits are better is I dont always have to wait for a reverse signal.. sometimes I can spot things happenin that the bot doesn't understand yet how to interpret.....
This strategy probably does not make much sense at all. Here is how some different scenario's can play out....
-Situation #1- we have a big trend day either up or down, but yet the bot losses money.. how can this be? Lets assume it is a big up trending day.
-ANS-
The whole day may be very positive and trending upwards. However the medium trend I track is showing some downward movement... this is probably due to a pullback during lunch or a common reversal time such as around 9:30.... central. So this trend is showing down... the small time frame is now starting to move up.... the bot enters a short position..... It sucks, but can happen sometimes.....
Other times this happens is when a BIG part of the move happens prior to 9:00 central when the bot is not allowed to trade yet. by the time the bot is allowed to trade the market's medium trend is generally a pullback against that move... thus we end up taking a short at some point.....
-solution-
Why don't I simply make sure the medium trend agree's with the longer trend? I have tried this and it really degraded the whole system. It seems a lot of money is made when the medium trend gets out of sync with the bigger part of the day...... Generally its seems that if the medium trend and longer trend both agree that the move against those 2 trends sometimes becomes rather large new trend.. again something like that could happen around key reversal times or during lunch.....
-Situation #2-
Market is very choppy, so the bot gets killed, why?
-Ans-
This one is pretty obvious... what happens is the market is choppy and doesn't often revert back to the medium trend quickly.. instead the medium trend normally soon takes the same direction as the short trend... as soon as these 2 come in sync with each other the short trend starts to move the opposite way so the bot initiates a trade against it and the same situation happens over and over..... and the bot gets killed....
After reading this, you guys may think I am completely nuts and my strategy is bogus... the only thing I know is that it backtests well and I think it will become profitable here in the short future again...... I am looking for some feedback on this type of strategy... perhaps some filters that may help identify good times for a strategy like this to trade.... I do not have time at the moment, but if requested i can pull together a few screenshots from some days showing where it enters to better demonstrate the type of strategy this is.....
I am looking for some constructive criticism here and hopefully by sharing the basics of the strategy some of you will be willing to offer some ideas on how to improve it....
Thus far it has been classified as a counter trend strategy..... I let this classification go as it is "sort'" of close. At least when you look at the positions it enters... The strategy is more of a... umm.. I guess lets call it a reversion to bigger trend strategy... Or I guess you can call it trading the pullbacks.....
The bot trades on a VERY small time interval... what it looks for is that on a slightly bigger time interval is an underlining trend. it wants this trend to be as strong as possible... this bigger trend is not on a much bigger time interval. only slightly bigger.... which leads to the possibility that my bigger time frame trend is AGAINST what the whole day's trend is like...
So now that a trend is established there, it looks for a trend going the opposite direction on the smaller time frame.... As long as the bigger time frame's trend is still intact we will enter a position which goes with the bigger time frame or in other words against what the short time frame is doing..... This helps a lot with slippage and actually normally gives me positive slippage... because the market is still aggressively moving against where I entered....
One drawback is normally my position is in the red at least for the first couple of bars... but it then generally starts doing as it should and reverting back to the medium trend....
We only exit when the opposite signal is given or if the market appears to be becoming choppy.... I generally use ADX here.. if the adx drops below <15 I exit the position.. this is my only exit besides a reversal signal...
one of the reasons I think my manual exits are better is I dont always have to wait for a reverse signal.. sometimes I can spot things happenin that the bot doesn't understand yet how to interpret.....
This strategy probably does not make much sense at all. Here is how some different scenario's can play out....
-Situation #1- we have a big trend day either up or down, but yet the bot losses money.. how can this be? Lets assume it is a big up trending day.
-ANS-
The whole day may be very positive and trending upwards. However the medium trend I track is showing some downward movement... this is probably due to a pullback during lunch or a common reversal time such as around 9:30.... central. So this trend is showing down... the small time frame is now starting to move up.... the bot enters a short position..... It sucks, but can happen sometimes.....
Other times this happens is when a BIG part of the move happens prior to 9:00 central when the bot is not allowed to trade yet. by the time the bot is allowed to trade the market's medium trend is generally a pullback against that move... thus we end up taking a short at some point.....
-solution-
Why don't I simply make sure the medium trend agree's with the longer trend? I have tried this and it really degraded the whole system. It seems a lot of money is made when the medium trend gets out of sync with the bigger part of the day...... Generally its seems that if the medium trend and longer trend both agree that the move against those 2 trends sometimes becomes rather large new trend.. again something like that could happen around key reversal times or during lunch.....
-Situation #2-
Market is very choppy, so the bot gets killed, why?
-Ans-
This one is pretty obvious... what happens is the market is choppy and doesn't often revert back to the medium trend quickly.. instead the medium trend normally soon takes the same direction as the short trend... as soon as these 2 come in sync with each other the short trend starts to move the opposite way so the bot initiates a trade against it and the same situation happens over and over..... and the bot gets killed....
After reading this, you guys may think I am completely nuts and my strategy is bogus... the only thing I know is that it backtests well and I think it will become profitable here in the short future again...... I am looking for some feedback on this type of strategy... perhaps some filters that may help identify good times for a strategy like this to trade.... I do not have time at the moment, but if requested i can pull together a few screenshots from some days showing where it enters to better demonstrate the type of strategy this is.....
I am looking for some constructive criticism here and hopefully by sharing the basics of the strategy some of you will be willing to offer some ideas on how to improve it....