Frosty's auto-trading bot goes live with REAL money

Discussion in 'Automated Trading' started by frostengine, Nov 14, 2006.

  1. I have been debating this for months, but today I finally did it. I put my auto trading bot live. Not only live, but fully unatended in my real money account. The bot is trading 1 ER2 contract at a time.

    The purpose of this thread will be two fold.

    #1 will be a place for me to record the bots results and get feedback on potential improvements. Also, it will give me motivation to KEEP running the bot even when it has bad loosing days. Historically the strategy has shown some severe draw downs that scared me from trading it. But everytime it proved it woudl have rebounded and made money in the long run. This here i am hoping will provide pressure to keep me turning it on everyday.....

    #2 will be to show that a fully auto trading strategy can be profitable. Hopefully it will provide motivation for other developers to continue their work and create good auto trading strategies.

    Because of my temptation to tamper with the bot's trading, I am leaving it running complety unattended. I do not want to touch it at all. I am very nervous about seeing what todays results will turn out to be. Especially considering yesterday on the sim account it lost $600..... I figured it had a bad day yesterday so the chances of a bad day again today historically has been low.... so fingers crossed....

    If needed to prove that that all results are legit I can post end of day screen shots if requested.
  2. WOAH! Great day.... The bot made 2 trades, first was a short trade this morning that it closed out even.. it stoped and reversed here and ended up gaining almost 15 ER2 points....Feeling REALLY good about choosing today to go live...

    Net for day: $1,450.40 on two trades....
  3. GS19


    Nice, keep posting
  4. Today was a nice surge rally all day. On days like this review the rules and make sure that they are designed to catch days like these and cut losses or get out when it more choppy. That will really push the results higher.

    Good luck!

  5. After a great start today, I feel very confident running in my live money account tomorrow. The big win today gives me a fairly big cushion to work with incase I run into a draw down period. However, historically wends has shown to be the #1 day of the week with tuesday being the #2 day of the week. So I think the odds are on my side tomorrow. I wouldn't be surprised if the bot has a down day though.

    Does anyone use day of the week consitantly to filter which days the program will trade? For example over the past 3 months, tuesday and wends added together account for nearly 70% of all the profits.

    Thursdays are the third best day accounting for 13% of the total...

    Monday's on the other hand are VERY dismal, accounting for only 9% of the total, but at least its a still a profitable day....

    Friday is also a pretty bad day comming in around 9% as well.

    It would seem to make the most sense to only trade on wends and tues, but everyday has a positive expectancy so its a little hard to just decide not to trade on those days and throw away that money. Although, the stress of seeing losses could be avoided greatly by only trading those days. I am unsure why those 2 days seem to do so much better. Perhaps more economic news comes out those days so the market is more trendy, or the ranges are bigger?

    One use of this knowledge could be to decrease size on days besides tues and wends. On tues and wends INCREASE size.... this could help create much bigger returns. Something I will have to look into.

    Can't wait for tomorrow to see what it brings.
  6. Nicely done frost. I know switching a bot from the paper account to the live account is one of the most stressful parts of my week. It held on long enough for 15 points? I generally work with much shorter time horizons. Do you just use a very big trailing stop or something that looks at market conditions which decides if it should bail on the position?
  7. Another day in the books. Today the bot made $40 after comish. No where near yesterdays great day, but in the green. So very pleased with performance thus far. Yesterday was way out of the norm from historical testing, but within limits of past events. Today was a more realistic day. I checked on the bot's progress about 15 minutes before the fed minutes and the bot was up $410 from a long position it had opened much earlier in the day. After the fed minutes the bot proceeded to give up nearly all of that trade's gains before finally reversing and going short. The short trade did not get the confirmation needed to continue and exited pretty quickly for a 2 tick gain.

    I wanted so badly before the fed minutes to close that position to keep the $400 because I know the bot has problems with sudden movements sometimes. I did not close it however, due to the fact that I trust the bot's strategy and in the long run it will do just fine if I dont meddle in it's affairs.


    The bot trades on a VERY short time frame.... however the bot is designed to only exit a trade if the opposing signal is generated at which point it will reverse, or if not enough follow through is generated after an entry signal which happens when the market is entering a trendless phase... That exit is what happened after the reverse to a short position this afternoon. It detected not enough follow through and became undecided on which way the market would go. So although the bot is trading a very short time frame, the way its exits are designed it will ride a big move until that move is basically over.... One downside to this is it can be slow to react to VERY sudden moves like when some news releases come out or something happens that shakes the markets...

    Days like today though when I see it could have had a NICE profit makes me want to tinker with it and put some protections in place to protect profits. However, every trailing stop, profit target, and hard stop I have ever tested with it has shown a GREAT decrease in the expectancy and profitabiltiy of the system.
  8. Suprising that stop strats would reduce the profitability drastically. maybe the problem is with how you define profitability. Rather than looking at straight amount of money earned each day, it might be a better idea to gauge it as profit/(capital used in each position * sum of time in positions). that might give you a better gauge of how hard each dollar of capital is working for you. and is more realistic if you want to expand into multiple contracts / securities.
  9. Just looked back at some of my testing data. A trailing stop decreases the expectancy to only 1/5th of what it is without one, and takes a big hit on the reward/risk as well as decreasing overall profit by about half.

    All hard stops decreases expectancy by only about 20 dollars per trade, but it also lowers the reward/risk..... which isnt a good thing. overall total profits remain about the same... if the stop is lower than 7 ticks it creates a negative expectancy.

    Profit targets also decreases expectancy and the reward/risk ratio.... keeps the total profit fairly similar however..... profit targets does show to make the equity curve slightly smoother..

    It seems after everything that having good reasons why it gets out of a position or reverses is much better than any of these other exit methods...... may need to try some more exit strategies... just not sure which to try next.... its all basically icing on the top though since the method is very profitable as is.... its just hard to resist the erge to tweak it more.. especially when the exit is very un optimized compared to the quality of the entry singal....
    #10     Nov 15, 2006