Frosty's trading bot goes live part 2
- Thread starter frostengine
- Start date
OK, OK.
Anyone tak'in bets as to how long Frost can go, experiencing heavy loss after heavy loss and uncertainty about the probability of the next trade succeeding (not to mention understanding what the potential is for achieving defined profit targets) before he proceeds to turn the "live" portion of the bot off, only to see it rebound and start to make money again?
Any takers?
Oh, right, he's already done it.
That happens because while he has done the general backtest for the system, the truth is he can't psychologically handle the drawdown, so what he actually needs to do is design a system that he can trade.
This way he can weather the storm when the system goes into drawdown. As it stands, just as the bot is about to recover and begin to make incremental gains (the market moves in cycles, anybody remember me say'in that?
) he turns it off. :eek: Not the way to make money in the markets.
... and one more thing.
As the trendiness of the market decreases and its volatility increases that no stop loss policy has the potential for blowing through the bottom of his account (oh yeah, how much did he say he was trading with again - to take a $1,200 hit like he did, it would need to be at least, oh 24,000, most traders wouldn't take a hit like that with an account less than $48,000, and the old school types ... you guessed it $120,000).
Now, I know what I just wrote isn't going to make sense to WJ and the rest of his crew, but that's because they never bothered to actually learn, well, anything.
Happy 4th,
Jimmy Jam
Anyone tak'in bets as to how long Frost can go, experiencing heavy loss after heavy loss and uncertainty about the probability of the next trade succeeding (not to mention understanding what the potential is for achieving defined profit targets) before he proceeds to turn the "live" portion of the bot off, only to see it rebound and start to make money again?
Any takers?
Oh, right, he's already done it.
That happens because while he has done the general backtest for the system, the truth is he can't psychologically handle the drawdown, so what he actually needs to do is design a system that he can trade.
This way he can weather the storm when the system goes into drawdown. As it stands, just as the bot is about to recover and begin to make incremental gains (the market moves in cycles, anybody remember me say'in that?
) he turns it off. :eek: Not the way to make money in the markets.... and one more thing.
As the trendiness of the market decreases and its volatility increases that no stop loss policy has the potential for blowing through the bottom of his account (oh yeah, how much did he say he was trading with again - to take a $1,200 hit like he did, it would need to be at least, oh 24,000, most traders wouldn't take a hit like that with an account less than $48,000, and the old school types ... you guessed it $120,000).
Now, I know what I just wrote isn't going to make sense to WJ and the rest of his crew, but that's because they never bothered to actually learn, well, anything.
Happy 4th,
Jimmy Jam
frost,
i did not follow this thread and actually do not intend to.
yet i ran quickly over some pages and get the picture
things are not going that well. my advice:
1. identify what the flaws were/are. by that i mean
it can be basically one of two things: trading cost under-
estimation or overfit. it should be relatively easy to check
for the first. look at all your entries versus signal price
and see if the difference fits with your backtest assumptions.
for the second check if you find that current conditions
are typical such where your system should make or
lose money or be break even. by current conditions i
mean range, vola, trendiness. if you find that current
conditions are accidentally bad, worry as little as possible
and continue the game. if it is not like that start from
scratch.
2. if you can make a living from such threads go on
with it, you obviously attract attention. if you are
serious about trading, stop being so exhibitionistic.
does not pay bills. i understood why you started it in
the first place, but i would think it distracts you more
than necessary now.
do well.
man
i did not follow this thread and actually do not intend to.
yet i ran quickly over some pages and get the picture
things are not going that well. my advice:
1. identify what the flaws were/are. by that i mean
it can be basically one of two things: trading cost under-
estimation or overfit. it should be relatively easy to check
for the first. look at all your entries versus signal price
and see if the difference fits with your backtest assumptions.
for the second check if you find that current conditions
are typical such where your system should make or
lose money or be break even. by current conditions i
mean range, vola, trendiness. if you find that current
conditions are accidentally bad, worry as little as possible
and continue the game. if it is not like that start from
scratch.
2. if you can make a living from such threads go on
with it, you obviously attract attention. if you are
serious about trading, stop being so exhibitionistic.
does not pay bills. i understood why you started it in
the first place, but i would think it distracts you more
than necessary now.
do well.
man
this sums up your ignorance on the subject of automated trading and specifically frost's bot. its very surprising that you've been following this thread since part 1 and still seem to know nothing about the situation. frost had every possible reason to go live with his new system, he had a well tested bot which had a very nice equity curve and statistically backtested support.
when making the jump from papertrading to live in any system, hand traded or automated, having a backtested + papertested / statistically proven method is the most you can hope for as a sign that it will work live. so of course he was right to go live with his new system.
supposed to have issues ironed out? oh please enlighten us on what 'issues' have not been ironed out? the bot seems to be functioning exactly as programmed, the live results are matching the paper results. the only thing that isn't peachy is the fact it is in a drawdown phase. how exactly does one 'iron out' that? id love to know. if the bot made 10k before entering this same drawdown phase, we wouldn't be having this conversation, since its paying with 'house money' in your eyes, statistically speaking there is no difference.
sure drawdowns suck, losing money is part of the game. but if you have NO reason to think that your statistics are incorrect (like frost, paper trading still performing the same as live, and paper trading did perform as backtested before), its a price you have to pay. my original system lost money out of the gate and if i was unlucky enough to come across your 'advice' before the statistics in my system kicked back in i might not be trading a profitable system today. so thank god i didn't read any of your posts while i was starting out.
am i saying that frost's bot is going to become profitable on the live account? no. do i think that it will continue to perform the same as the paper account? yes. there is a chance the system has become invalid or has entered a long period of unprofitably. but the last year+ of testing has given frost the statistics which he is currently betting on. throwing that all away and saying its wrong to have started running live in the first place is just plain stupid.
uhh. do you even listen to yourself anymore? maybe he would be able to handle the drawdowns a bit better if YOU weren't telling him to scrap the bot every down day without any reason other than you feel the losses are too big. luckily i think frost is smart enough to ignore you and not let your ignorance influence his decision making process. im sure if he has a real reason to scrap his bot he will, until then, like the rest of us, hes betting on the statistics from testing and the validity of the test environment.
personally i dont think its that bad of a bet. if his year+ of work is flawed he loses some money, if it wasn't, he and his children might be set for life. i made a similar bet when going live and its starting to pay off slowly. i agree with his choice to run the bot back on the paper account only to prove that the system is still valid in the emerging market.
anyways i am enjoying this conversation since these are the same issues iv had to deal with and still am dealing with when trying to push an automated system onto the first step of profitability.
"having a backtested + papertested / statistically proven method is the most you can hope for as a sign that it will work live"
Crucial is that temporal unfavorable market conditions do not shake you out of confidence when going real. That is, one should avoid losing too much too soon while phasing in the strategy. You can usually achieve this by lowering bets - this is problematic in case of futures because sometimes a single contract is too much. Another considerable way of scaling down is that you re-tune strategy parameters that it trades more safely. But this can result in lower expectancy.
A solution to consider (just an ad-hoc idea): you could toss a coin before entering a trade, if head, take the trade, if tail, skip the trade. With this method, you get over a potentially bad market phase after start with a lower drawdown.
Crucial is that temporal unfavorable market conditions do not shake you out of confidence when going real. That is, one should avoid losing too much too soon while phasing in the strategy. You can usually achieve this by lowering bets - this is problematic in case of futures because sometimes a single contract is too much. Another considerable way of scaling down is that you re-tune strategy parameters that it trades more safely. But this can result in lower expectancy.
A solution to consider (just an ad-hoc idea): you could toss a coin before entering a trade, if head, take the trade, if tail, skip the trade. With this method, you get over a potentially bad market phase after start with a lower drawdown.
Frost is actually a very talented programmer to even have gotten this far in the game.
Another idea of how he could make money is to actually teach traders how to design trading bots (for a nice fee of course). I am sure he could make a substantial amount of money doing it, and the market would most likely increase as traders became aware of him and his services.
As austinp once said, "There are a lot of ways to make money in this game, and trading is just one of them".
I'm coming so hard because these are the lessons I've learned from hundreds of hours of screen time and multiple trial-and-error experiences (some very painful, quite frankly) on the road to becoming a consistently profitable discretionary trader with the ability to scale-up in contract size and make some serious money.
If what I'm writing doesn't work for anyone, fine. But the market is much much harsher, and the ultimate in an impersonal taskmaster. We work so hard at this because, for most if not all of us, working for a living just is not an option.
Good trading,
Jimmy Jam
P.S. Here's a guy who has designed bots for public consumption: Thinking Stuff
Another idea of how he could make money is to actually teach traders how to design trading bots (for a nice fee of course). I am sure he could make a substantial amount of money doing it, and the market would most likely increase as traders became aware of him and his services.
As austinp once said, "There are a lot of ways to make money in this game, and trading is just one of them".
I'm coming so hard because these are the lessons I've learned from hundreds of hours of screen time and multiple trial-and-error experiences (some very painful, quite frankly) on the road to becoming a consistently profitable discretionary trader with the ability to scale-up in contract size and make some serious money.
If what I'm writing doesn't work for anyone, fine. But the market is much much harsher, and the ultimate in an impersonal taskmaster. We work so hard at this because, for most if not all of us, working for a living just is not an option.
Good trading,
Jimmy Jam
P.S. Here's a guy who has designed bots for public consumption: Thinking Stuff