Just wondering if someone could point me to some papers/books that might help me develop my personal algo trading system. I am planning to start small (<$25k), and am not looking to hold positions typically beyond a few days - ideally, I wouldn't want to carry anything overnight. I am new to quant finance, and though I possess a bit of knowledge of the markets, I have never traded before. So right now, I am just going to focus on data gathering, model development, backtesting, etc. as I continue learning about the markets. I have a Machine Learning background , so anything from that domain will be OK as well.

Since you say that you have "never traded before" I would perhaps suggest that you do some basic trading first, maybe with virtual money. Although it seems like starting from the buttom, my personal experience indeed promotes this procedure. Also, I don't know how much mathematical/econometrical background you have (please elaborate). If having some, you will quickly pick up some feature on which you can apply your models.

QuoteOriginally posted by: volatilityManSince you say that you have "never traded before" I would perhaps suggest that you do some basic trading first, maybe with virtual money. Although it seems like starting from the buttom, my personal experience indeed promotes this procedure. Thanks for your note. Yes, I wasn't going to start trading right away. I was planning on only building models and backtesting for now. QuoteAlso, I don't know how much mathematical/econometrical background you have (please elaborate). If having some, you will quickly pick up some feature on which you can apply your models.I have done ML work, so have felicity in the Mathematical/Statistical paraphernalia associated with it - Calculus, Linear Algebra, Probability, Stochastic Processes, Statistical Inference, etc. But that really is not a constraint. I can pick up from papers in other areas of Math or Econ as well. I am currently reading Ernie Chan's book, and am also looking through Wilmott's and Mark Joshi's books. I find the latter two to be more academic than practical which is not what I am looking for right now. I am also leafing through some general books like Larry Harris' Market Microstructure, Bodie and Kane's Investments, and 1 portfolio optimization book whose name eludes me this instant. A good set of practical books and/or papers with a focus on short-term/intraday trading would be helpful. A lot of the algorithms I am coming across seem to be biased towards the 'buy and hold' strategy.

Just a philosophical comment as you search for reading. The markets are quite efficient and there is everyincentive for them to become more so over time. So, what tends to happen is that traders discoveranomalies/pockets of inefficiency, and get to exploit them for a certain time and then competition,rule changes, law changes, whatever, dilutes the thing. Given that, think of yourreading search as a search for stories about historical inefficiencies.Examples:"Beat the Market" Thorp & Kassouf (pre CBOE warrant market)ex-dividend arbitrage in the listed options market (more recent example)"Flash Boys" (co-location exploits)....Try to assemble your own list. Ideally, what would happen would be that traders latch on to something,exploit it, it fades, they retire to the beach and write a book about it, clearly explaining why it disappeared.In practice, the "end-game" stories are not so clean-cut as this, so you should approach the existent stories quite skeptically.

Alan's very nice post reminded me of a published anecdote about Fischer Black. When he was in academia Black did a lot of thinking about the pricing of futures contracts, especially financial futures, which had recently been introduced. When he went to work for GS he began to take action based on his theoretical knowledge. Noting that some stock index futures were in backwardation (far contract priced cheaper than nearby contract) and that this was in contradiction with [$]F=Se^{(r-d)T}[$] he began to buy distant futures, hedge them and hold them until delivery. Soon he was able to make quite a bit of money, but the markets changed and began to price the futures correctly, i.e. in contango. Other people had begun to catch on. The profitable opportunity was gone forever. Today that equation is taught in the first class in derivatives pricing and for that very reason is not useful any more.What does it tell me:0. F Black was a very smart man1. You *CAN* make money from theory, but not by directly applying the theory as expounded in Wilmott, BodieKane & Marcus, and other books. Rather you have to understand what the theory says must hold true and then find real-world examples where it does not hold and take advantage of that. You have to learn the theory thoroughly and then close the book, get out of the library and go look in the real world for what Alan calls empirical anomalies that are exploitable. The second activity takes a completely different mind set from the first, and requires access to live data.2. The anomalies end after a while. They are more common in new markets, but markets become more efficient over time. When the game ends you have to start over.3. For these reasons trading is best learned in an apprentice fashion, working with other successful traders to learn what other people are doing ("how does GS make its money nowadays?") what kind of approaches and data people use in what markets and so on. Once you have a feel for this you can try to apply your knowledge of Statistical Learning or whatever to systematize/automate or improve upon what people are doing. But just trying to apply powerful mathematical techniques on a blackboard in your office is not useful I think. You have to be immersed in the trading world. Trading is more like a game of trying to outsmart the next fellow than a business or profession to which general techniques are applicable.

have a look at the following:https://www.quantstart.com/ebookshttps: ... ding-ListI guess you will find what you are looking for here

My main rule is to stay from all books with any of the following words in the title: Finance, Quant, Trading, AlgorithmStick with the best statistics and machine learning books. One exception:

QuoteOriginally posted by: AlanJust a philosophical comment as you search for reading. The markets are quite efficient and there is everyincentive for them to become more so over time. So, what tends to happen is that traders discoveranomalies/pockets of inefficiency, and get to exploit them for a certain time and then competition,rule changes, law changes, whatever, dilutes the thing. Given that, think of yourreading search as a search for stories about historical inefficiencies.Examples:"Beat the Market" Thorp & Kassouf (pre CBOE warrant market)ex-dividend arbitrage in the listed options market (more recent example)"Flash Boys" (co-location exploits)....Try to assemble your own list. Ideally, what would happen would be that traders latch on to something,exploit it, it fades, they retire to the beach and write a book about it, clearly explaining why it disappeared.In practice, the "end-game" stories are not so clean-cut as this, so you should approach the existent stories quite skeptically.Good idea, thanks. It might not be easy to find such literature though. I did have Thorp, Michael Lewis, and a few other popular authors on my radar. Perhaps I'll skim through those books as well. I have been thinking of crawling websites of finance/econ journals to get access to their historical table of contents data (mostly title and abstract of papers) and generate a timeline of their themes, mine for keywords etc. We'll see how far that would take me.

QuoteOriginally posted by: acastaldoAlan's very nice post reminded me of a published anecdote about Fischer Black. When he was in academia Black did a lot of thinking about the pricing of futures contracts, especially financial futures, which had recently been introduced. When he went to work for GS he began to take action based on his theoretical knowledge. Noting that some stock index futures were in backwardation (far contract priced cheaper than nearby contract) and that this was in contradiction with [$]F=Se^{(r-d)T}[$] he began to buy distant futures, hedge them and hold them until delivery. Soon he was able to make quite a bit of money, but the markets changed and began to price the futures correctly, i.e. in contango. Other people had begun to catch on. The profitable opportunity was gone forever. Today that equation is taught in the first class in derivatives pricing and for that very reason is not useful any more.What does it tell me:0. F Black was a very smart man1. You *CAN* make money from theory, but not by directly applying the theory as expounded in Wilmott, BodieKane & Marcus, and other books. Rather you have to understand what the theory says must hold true and then find real-world examples where it does not hold and take advantage of that. You have to learn the theory thoroughly and then close the book, get out of the library and go look in the real world for what Alan calls empirical anomalies that are exploitable. The second activity takes a completely different mind set from the first, and requires access to live data.2. The anomalies end after a while. They are more common in new markets, but markets become more efficient over time. When the game ends you have to start over.3. For these reasons trading is best learned in an apprentice fashion, working with other successful traders to learn what other people are doing ("how does GS make its money nowadays?") what kind of approaches and data people use in what markets and so on. Once you have a feel for this you can try to apply your knowledge of Statistical Learning or whatever to systematize/automate or improve upon what people are doing. But just trying to apply powerful mathematical techniques on a blackboard in your office is not useful I think. You have to be immersed in the trading world. Trading is more like a game of trying to outsmart the next fellow than a business or profession to which general techniques are applicable.Yes, ideally you would want to have someone providing you personalized advice but alas, I have no such luxury. I did get an offer with a small brokerage house to serve as a part-time consultant for the quant analytics platform they're building, but I did not follow it up because of a few academic commitments. In hindsight, that might have given me an opportunity to form a network. It's not GS, but it will be a start I guess. Maybe I'll ping them again and see if they're interested. Thanks for the idea !

QuoteOriginally posted by: algoisgamehave a look at the following:https://www.quantstart.com/ebookshttps: ... ding-ListI guess you will find what you are looking for here :)That's a big list, thanks !

QuoteOriginally posted by: AnalyticalVegaMy main rule is to stay from all books with any of the following words in the title: Finance, Quant, Trading, AlgorithmStick with the best statistics and machine learning books. One exception: actually that's the approach I will take in the longer term. I usually don't worry too much about the methods others use, and just try and play to my strengths. But since this is a fairly mature field, and people have been at it for a while, I am interested in taking a closer look at what's been done, especially if I am able to find implementations of popular algorithms.

Alan has a very good point. There is IMO no such thing as "A strategy" (singularis). It's like a battlefield where the war changes all the time due to different circumstances. I have a good book which I think you should read, although it has absolutely nothing to do with financial markets. Instead it deals with war-strategies: read it, accomplished some trading experience as well, you'll quickly see some of the similarities between war and trading, in which you can develop your own models. Good read

QuoteOriginally posted by: roundandroundQuoteOriginally posted by: AnalyticalVegaMy main rule is to stay from all books with any of the following words in the title: Finance, Quant, Trading, AlgorithmStick with the best statistics and machine learning books. One exception: actually that's the approach I will take in the longer term. I usually don't worry too much about the methods others use, and just try and play to my strengths. But since this is a fairly mature field, and people have been at it for a while, I am interested in taking a closer look at what's been done, especially if I am able to find implementations of popular algorithms.If so..then check out these: >> some decent infohttps://www.amazon.com/Way-Trade-Online-Video-C ... +the+trade >> Technical not quant but a few solid ideas for understanding basic thinking in trading. >> describes a basic trading strategy you can build on.

QuoteOriginally posted by: AnalyticalVegaQuoteOriginally posted by: roundandroundQuoteOriginally posted by: AnalyticalVegaMy main rule is to stay from all books with any of the following words in the title: Finance, Quant, Trading, AlgorithmStick with the best statistics and machine learning books. One exception: actually that's the approach I will take in the longer term. I usually don't worry too much about the methods others use, and just try and play to my strengths. But since this is a fairly mature field, and people have been at it for a while, I am interested in taking a closer look at what's been done, especially if I am able to find implementations of popular algorithms.If so..then check out these: >> some decent infohttps://www.amazon.com/Way-Trade-Online-Video-C ... +the+trade >> Technical not quant but a few solid ideas for understanding basic thinking in trading. >> describes a basic trading strategy you can build on. an interesting set of books. I must admit I had never heard of any of them before. I'll take a closer look. Thanks a lot.

Look at https://www.quantstart.com
At a minimum get a backtester coded and working first.
Transaction costs could kill you if you're planning on HFT.