Some people know that it often comes back shortly. But the secret is that sometimes it comes back after a long time-span like 2 weeks.
Above is my software i am currently finishing testing... it works on longs and shorts.. here is the trade sheet... allow for margin of error given that this version is in TD i am now doing the forward testing.. IF you think this is cherry picked.. give me one or two stocks, or futures, and i will run and put up the results of the back test DO understand that TD is flakey in terms of pricing entry and that the results are only close the code is being tested with only 100 share basis... IF it works half as well forwards, which remains to be seen it will give me a relatively low risk (100 shares is not much), with a decent pay out I think it would be interesting to run this on others suggestions.. totally sure it will hold up... and totally sure it will stay a secret... the other version is written in python
mr artfldgr, you have made an offer i cannot refuse. Very kind and thanks for the oppt'y. How's about Feb Gold Futures contract? Cheers!
Ah... well the system is not predictive... it doesn't look forward... it determines what condition its in... which is why it gives by and sell signals.. ie at the moment... However, a promise is a promise and here is what the past year would have been like using it (as far as i can tell given the caveats) the settings are one year to last trading day, daily, 100 units (are they shares or contracts? sorry, wasn't sure of the nomenclature) the closest thing to a prediction it can give is the current state.. whether its still in a bear state or a bull state... i will start with that... About 20 days ago, it said buy to open, and has not given a close yet there is no buy short on the other screen (you will see them below) this is NOT a prediction if the next bar is what it requires, it will issue a sell if not, it will consider itself still in its up trend. ALSO, from studying this so far, its end recommendation is not as solid as its opening move. the span it uses when it opens a position is behind the open, and not while you have the purchase.. the point where it says sell, is after the turn, and so it gives back at the end, sometimes quite a bit if the turn is quick and hard... in essence it makes its cash in the back trading because of the run up or down is longer than the waste it gives up at the end of the run. this is the first iteration of this and i am still exploring why it works so well, which is needed to be able to improve it or mitigate its weaknesses... there are some buy and sell signals candle to candle... however, given that the market runs in trends longer than a 1 day reversal, these tend to be annoyances that i think will have to be accepted as tiny loss vs large gain from larger runs.. at least till i figure out some math to add (which i might not figure out) with that said, here is the bull run (no pun intended).. and here is the series of trade over the year... unlike other things it starts with a pretty large amount i am unfamiliar with trading gold futures according to the trades below the first trade using 100, is (to me) a whopping $224,000.00 with the final bull total being $3,912,000.00 here is the list of its trades and now for the bear side of the coin (the system trades both, so it doesn't matter if its a bull or bear market technically).. according to the sheet, its first trade cost $272,000.00 (to me, thats ouch) Total P/L: $1,889,000.00 given its an up market, the bulls make more than the bears.. if the market turned around, it would be the other way, so its balanced here is the list of trades someone that i talked to complained that it was only making a few trades but then i realized that was ok, given that it never left the market once its in meaning that when its not in a bull trade its in a bear trade oscillating, so its in all the time (im still learning how it behaves as i said.. and still amazed and gobsmacked) $224,000.00 $272,000.00 ------------------ $496,000.00 i guess the above is its initial investment.. though technically if its oscillating between bull and bear state you would only need the first one, as it would be closed, and then open the alternate position. $3,912,000.00 $1,889,000.00 ---------------- $5,801,000.00 total profit.. thats the run... sorry its not a predictive engine.. but given the results i dont think thats much to mind... i hope this was at least entertaining..
Mr Artfldgr, Good job avoiding whipsaws during the 6/25 to 8/1 timeframe as currently set up. Looks good to me, my man. Now for some good news on trade exit strategies you turn up. Cheers
Artfldgr Does your computer run shown trade the way the original poster describes his method?? or is your computer run a different independent method of trading. unfortunately the chart has so much words on it that it is impossible to see the trades. you would need to spread out the bars or take one trade at a time on a chart. it is impossible to see your trendlines where they begin and end. if those are trendlines used to make trades. can you comment on what you method does. it looks real nice. if you changed your trading unit from 100 to 1. then it would be per contract. right now you are trading 100 contracts of 100 ounces. or 10,000 ounces per trade. which would require $5 to $10 million dollar account. kind of a big account for most people. hahaha.
Well, it doesn't show trend lines... it computes a state and says buy... so its less about graphing for a person to make a judgement and more for a machine to have a finite condition of yes or no... so unless your watching it, or somehow searching for the condition on the last candle, your stuck...a machine would scan for the condition and start its trading series upon what it then finds as the last candle that says either sell to entry (short) or buy to enter (long)... then once its in, it never stops as there is no condition to get out... i originally thought this to be bad, but not really... given the outcome as long as a company is in business, and its not swinging like a crazy volatile wack job there is cash to get, and the code does that... especially if the runs are long... Given your so polite and this is exploratory inquiry, not claiming to rock the world, but trying to understand WHY this out of all the freaking things out there works so well... its unbeleivable and quite different... after all, most systems have in and out periods, not a constant short, long oscillation that lasts the lifetime of the firm once your in.. so almost everything that i am working with is having problems because of that situation... here are the charts again, with one contract (curious to see if small works anyway) and in multiple images so you can see the trades... i wont repost the trade sheet, as that is the same, just the amounts are different, but i will repost multiples of the screen so that you can look at what its doing and with a run of 1.. going to take me about 10 minutes (not to run, but to copy, crop, save and upload). thanks for the request... it keeps me from being accused of cherry picking.. i guess no one can believe that someone can come up with a different trading system (continous) that actually works very well... i can barely believe it.. but i can tell that its working despite how crappy TD is on this... (that is the entry or exit exact price value varies but not enough to undo the results, but more like if you were actually trading the exact moment you hit buy the price would vary without a limit or some other methodology)
these images should be easy to see the point of each trade, and if they are off a bit on entry (a TD limitation) its not much.. later when i get to run this on a better modeling system, which i am working on doing right away, there will be more accurate data (and in my other thread i started for this discussion). https://www.elitetrader.com/et/thre...algorithm-and-a-strategy.339781/#post-5001167 Also, if i can find a platform to trade it, i am willing to try it out with a small amount and go from there... if it works as well as it seems, you only need a small amount to start and then parlay from there anyway.. I decided to also post the trade record... would you trade leaving lots of scraps for a trading system that nearly always ended positive? i would... because then your really talking about risk, and if this works this way (and i know the math so i know it will -but always has to be proven), its risk is actually very low... its results are not with stop losses or anything protective... in fact you will note that in many cases it goes below the price you enter, but still comes out! if 9 out of 10 of the companies you put it on, it works out, does it matter what it leaves behind? what if it bats 10 for 10, and what varies is the amount only? bad companies it makes a little, great companies it makes a lot? that would be very different (and i am NOT claiming this) if its perpetual once it enters, does it matter that it can only trade a set, and not re-balance? the program is forcing me to ask some very interesting questions about the very nature of trading (though if this moves to the other thread, all the discussions will be encapsulated in a single location without distraction). many of our concepts of trading accuracy and risk in practice answer those questions differently!!! and we are very used to those answers always being the same we are loath to even question that they could be different. This is an interesting concept.. but why would a system that is different lead to the same practices for other systems? And you can see where TD is messing up... it really doesnt change things a huge amount.. and so a tighter model will be done and needs to be done... but the point i am making is that its right about when to get in, even if the price it gets in is sometimes off because of how TD allows one to enter the price (no market order).. ie. if you accept that its entry is right, and that its tracking a long run, then the mistakes are just going to vary how much it profited and rarely matter to whether it profited... which i am seeing a lot more charts than i could ever put up and been looking at that point very hard as it will be my money on the line (and i am currently unemployed and not finding much so its critical this works for me) anyway... here is the trade sheets first then the charts just for you being reasonable and helping me explore this IDEA another note.. in the strategy sheet (looking at what TD calls it now), you can see that its not 100% in all its trades (thats nuts)... some it does lose.. but its wins are always better than its losses, so i am just going to accept that.. Technically if this was trading as one unit, there would be only one entry not two so the one entry would yield a bull total and bear total that would be combined at the end given how TD is, and how i program it, i have it doing separate as its auto trade which should reverse and it doesnt reverse as it should in real life... ie. a machine or person could hit a reverse button and in that candle have it happen... not so well with td... TD does not expect a continuous system... no one does actually.. (and that was not what i was thinking about nor going for when i was designing the code to go after what it goes after!!!!) entry cost $1,306.00 $1,321.70 ---------- $2,627.70 entry cost.. Results summary $34,960.00 - BULL $18,890.00 - BEAR ------------------ $53,850.00 - Total and now the bear charts.. the tiny blue lines are TD way of saying your in the trade... we all know that some of the entry prices are off slightly, a quirk that future better models on another system will remedy... though you will note that they do not matter all that much to the outcome, but only to the amount which is each trade added up... one could (as i have in doing some basic looking) tweak those values in the strategy sheet to a bit more accurate and know this is true rather than assume one side or another based on hunch or belief. it leaves a lot on the table... but i think its doing so in favor of a better stable winning outcome, and that min max may actually do that... i would rather have a system that defines win and losing by how much it wins than one that actually wins and loses after an arbitrary period (that is long enough to have enough trades).. it would have been a good system when commissions were 40 for a trade (20 in, 20 out).. as its number of trades in and out if combined is quite low... let me know if you notice something i am missing that is not covered above. and try to move this to the other thread so that its not buried and lost...