Why publish an algorithm ?
- This topic has 11 replies, 6 voices, and was last updated 3 weeks ago by
.
-
There are many who don’t have an algorithm that regularly earns money. It’s not that easy. Here, people help each other to create ideas for algorithms that could work. Existing, tested algorithms are offered for sale on the marketplace to give people the opportunity to play on the stock market.
3 users thanked author for this post.
If you genuinely have an algo that works why not just trade it? the is a real questions I am honestly confused about the concept.
You are completely correct;
I don’t see a reason to make my systems public (read : license them for money) because those systems will earn the money in another way. And “helping” someone while it still could lose his/her money is not my thing.I emphasized one word in the quote from your text …
The context of the concept was (!) that people helped each other indeed. This is past history because the principles of the Market Place killed that concept (systems on the Market Place may even provide codes “stolen” from those aiding codes in the forum from ever back better times). So IMHO the concept failed but also destroyed a mere cosy community.
The subject has been passed before and this is not a complaint as such. But now you’ll understand your wondering better ?
Regards,
Peter1 user thanked author for this post.
Hi Morpheus,
first of all – late welcome to this community from my side as well. I was reading all your posts, you are putting lot of interesting and important questions, and receiving lot of very good answers. I wish and hope that with your specific background (as you mentioned blackjack and ig sprints…) you will progress very fast.
Regading concept of publishing the algos in case one has genuine edge. Did you see any such algo/code published? I did not. Ok, I might be blind. But I never saw one. On other hand I saw some guys here in the forum and and a bit more otside this forum publishing for my understanding quite interesting trading scenarios – at least worth coding yourself and testing. But these are very few, extremely few. And that is no surprise. The edges are extremely scarce. I assume they are much more scarce than majority probably can think of. IG states (as many other CFD brokers which are forced to publish such warning) that something like 75% of their customers lose money trading. For my knowledge it’s quarterly data/statistics. So after a bit oversimplified calculation one can tell that after 4 quarters / one year trading 99.5% accounts are losing. And regarding those 0.5% which are not losing: it does not mean they are earning consistently, it does not mean they have an edge at all…
Looking at the number of systems offered for purchase publicly one might think: so many edges, so many ways to earn money trading… I am quite sure that probably the only „edge“ those system-sellers have is that there are people who naively believe that system-sellers have detected some edge. This is my reasoning why system-sellers sell bots:
- They don’t have any edge, but as everybody, need some income (if not inherited millions) and look for people paying them for their beautiful curve-fitted backtests -> I assume it’s most probable reason
- They have detected possibly an edge, but
a) Don’t have enough capital to run those systems on sufficient level in order to make living, or
b) Don’t want to make living from trading
b: I could understand as a reason, because „normal“ people don’t like the volatile way of income generated by trading. It’s quite sure more comfortable to earn steady 2-3-5k every month by selling bots (and books and webinars and programming services etc) compared to trading where you might lose 15k one day, gain 20k next day etc and so on, just in order to make some 10k end of the month – and that not for sure, and not every month.
Regarding a: and here I am talking possibly some kind of solution:
I mean, if one has a system or in best case couple of them with a serious edge, but not enough capital to run them, and in case one is serious about trading, then one should not sell bots for some peanut amounts of EUR,USD,GBP… one should rather look for investors, external capital, right? I am thinking of one company (and sure it’s not the only one in the world), originally from Spain but now located UK, which is on one hand very good broker, and on other hand they offer opportunity to attract external capital to your trading. So you trade via that company as a broker, and if you wish, you can publish your performance, make your trading „investible“, and in that case
- if your performance meets this broker’s standard criteria, they might allocate tons of their own money to your trading, and/or
- everybody (so „you and me“) from „outside“ can open an „investors account“ and also then everybody who likes your performance can invest any amount to your trading.
So if you have an edge proven by your performance you can trade millions which you don’t have but attract from others. And then you share profits (I think currently some 15% go to trader, others go to investors and the broker).
Very solid, honest concept I would say, not only for those who have not enough capital…
Ok, here comes the sobbering part: currently they have like 3000 investible „strategies“ from their traders, more than 100M of own and and external capital allocated to several of those strategies, BUT – to my judgment less than 5 strategies out of those 3k are doing well/kind of good over many years of the track record until now. So far about the scarcity of the edges in the financial markets…
Another „side effect“ for us as PRT/PRC users: with that broker you can’t use PRT/PRC if you trade automatically… so one would need other programming knowledge or contract a programmer, who translates PRC into other languages… and here I am myself a bit „paranoid“ – I don’t want to share my algos to any programmers 😀
Cheers
justisan
1 user thanked author for this post.
hi phoentzs,
great to hear, that you are doing much better than before! and sure you have to thank yourself for it – I was just “blabbering” but you were helping yourself with you open mind.
were you possibly thinking about that in the meantime: why simple things work in the markets, and complex do not (or not so well…)? I mean, in the end the reason does not matter – we are not interested in the reason, but in the result. Yet still, it’s simply interesting.
yep, I was staying away for some months from forum (and from PC in general), but not from my trading 😀 I had to manage some private challanges in spring, and then enjoyed long summer with my kids at various places in Europe. but trading was all the time running its way. in the meantime I moved all my major strategies from CFDs to futures, but all of them and some “not major” strategies with 10% of total capital I am still running on CFDs, it’s fun and very useful to observe how similar/different those strategies “behave” on CFD vs futures, it gave me in the past and gives me still some insight which I use for improvement of my algos, for futures and for CFDs trading as well, and as you might guess, “improvement” means in my case almost always kind of simplification.
justisan
1 user thanked author for this post.
hi Swedish trader
one would need real corresponding experience in order to answer your question. means, one should have had real/proven edge and have lost it after too many market participants applied that particular strategy. I think probably we have here in this forum nobody with that kind of experience. what I can tell is 2 things:
on one hand, I have been reading plenty of interviews with hedge fund mangers etc, and what they are telling is, that their edges were slowly but surely disappearing after some 10-15-20 years. which is kind of goods news, right? so many years – it’s long time. I believe as well, just from logic point of view, that serious edges do not disappear overnight, it probably takes long time and huge amount of capital chasing that particular edge.
on other hand, your algo seller’s arguments are, I assume, not fully true. with those kind of statements they want to calm you down – or they simply don’t know better. it’s true that if you “sign” an CFD on let’s say nasdaq, you are in fact not buying or selling nasdaq stocks or futures. but if CFD broker’s customers all of the suddent have many more long positions than short (or vice versa), serious CFD broker will hedge himself in the real market. and so, if volumes are big enough behind what CFD customers are doing, it will affect the real market over long time due to hedging actions of the broker and so it will influence the edge.
again, I don’t believe it’s easy to “kill” serious edge. but that gives no serious reason to spread your setups accross whole www.
the more probable scenario than that the edge is killed, is that an algo “stops working” simply because behind the setup there was no edge anyway…
cheers
justisan
1 user thanked author for this post.
algo “stops working” simply because behind the setup there was no edge anyway…
And / or the price / market action upon which an Algo was optimised is changing / has changed.
In my opinion, if we have a significant downturn / correction on major markets then many of our own algos and also those on MP would take a dive in performance.
We’ve had a 2, 4.5 or even 16 year uptrend (see attached for DJIA) depending on one’s time horizon. Many algos currently showing good performance are biased towards those uptrends, even if the algos do take Shorts.
Longs and Shorts need to behave behave different in a downtrend (than in an uptrend) to make good profits.
Just a few thoughts … to add to the already good discussion!
1 user thanked author for this post.
The simplest and most important thing is that the markets are mainly rising. And since I specifically only trade SP500 and Nasdaq, this is especially true. And since I have eliminated any short from my approaches and try to move with the markets, I have had significantly lower drawdowns and better performance. And then I use very simple things to determine the trend and build scalpers in M1 or mean reversal or trend followers in higher time frames.
2 users thanked author for this post.
… in fact the question about “edge” is extremelely, super tricky thing… how does one know, if one’s strategy has an edge at all? and how does one know, that the edge – assuming one has an edge at all – is going away, or is gone completely. honestly, I have know clue, no serious criteria known to me how can one know and define that all.
if one’s algo does not earn money recently as it did last let’s say 3 or for may sake 6 months or even 3 years, just because it’s long-only algo and now it’s not any more uptrend but downturn phase in the market by someone’s delibarate defintion, it’s not the same for me as “losing edge” – it’s kind of to be expected-behaviour of long-strategy. system with the edge should stop diving and resume earning again after the downturn is over. if one has to adjust parameters of strategy with every new “uptrend”/”downtrend” because otherwise system is not earning money, then I would rather assume system does not have edge anyway – it’s possibly just hyperoptimized for the current/recent situation and with it one will stay in never ending optimization to the recent conditions, with huge time lag behind current conditions. even the most crapy system will have its “run” for some time – does not mean it has or had an edge. so, on other hand – a system has possibly an edge, and only possibly, in case it worked for years during various phases of market dynamics – and so maybe, just maybe, one can expect it to work for another years in the future, during various phases of markets, does not matter if in “uptrend” or “downtrend” or consolidation-shmoncolidation. this is at least my approach, and therefore I write “uptrend”, “downtrend” with inverted commas, not meaning those terms too seriously – because I kind of do not know which phase the market is in right now and in which phase it will enter tomorrow, meaning – I do not make my strategies dependent on such conditions, my algos “do not know” if currently the market is in uptrend or downtrend, these algos are all total dumbass, they process all the time the same ultra-simple-ultra-stupid go-long and go-short signals every day, with no “filters” for trend-conditions behind. and basically only this total-dumbass-approach gives me slight hope, that those systems have a little edge and will possibly perform tomorrow as well.
have fun trading!
justisan
1 user thanked author for this post.
Find exclusive trading pro-tools on 
