Best Practices for Algorithmic Trading Systems

I am currently compiling a list of the most basic strategies used in algorithmic trading.

• Trend Following (+Momentum)
  • MA Cross
  • EMA Cross
  • MACD (ingpawat)
  • RSI / LRSI (Mike_Trdw)
  • New Low or High
  • ATR (Away-Box793)
  • VWAP
• Seasonal
  • Sell in May and Stay away
• Mean Reversion (Mike_Trdw)
  • Mean Reversion To Trend
  • Mean Reversion in Range (The-Goat-Trader)
  • Reverting Market (The-Goat-Trader)
• Momentum Rotation (Tactical Allocation) (The-Goat-Trader)
• Grid Trading (Mike_Trdw)
• Arbitrage
• Offset Trades / Trading Pairs
• Index fund rebalancing
• Market timing
• Scalping
• Price Pattern / Candle Stick
• Price Forecasting
  • Neural Networks
• News-based
• Market Sentiment
• Trend line
  • Break
  • Bounce
• Standard SMA
  • break (SMA 20D, 50D, 100D, 150D, 200D)
  • bounce
• Range Breakout
  • Open Range Break Out
  • Horizontal Compression Breakout
  • Wedge Compression Breakout
• Options
  • Covered calls (Chuyito)
  • "The Wheel" on LargeCaps (faancy5050)
  • Short Iron Condor on SPY 0DTEs (faancy5050)
  • Long Straddle or Long Strangle on any earnings (faancy5050)
• Smart Money Concepts (good read, Franco_Love)
• "Martingale" (reckless_homicide)
  • Me: It is risky but it is a classic and basic strategy for you to play with. There are good papers on it too, so it made the list.

---

• BT_2112 mentioned that most of those strategies can profit from applying those on multiple timeframes simultanously. Since I do this too when it comes to my manual trading, I can see how this can be a very good improvement/factor when it comes to algorithmic trading.
• EntrepreV Just dropped the content bomb on us. He mentioned the paper 151 Trading Strategies which is just 370 pages of 151 trading strategies explained including formulas and stuff. I have yet to start reading, but I guess, if you want to have some more strategies to choose from, this might take home the whole cake. Many thanks!
• disaster_story_69 Mentioned having a visit of Trading Views library of pine scripts. I took a look, and it looks like a ton of content one can mine, especially if the pine scripts open source.
  • Link to the library (filter = strategies):
  • Link to the editor's picks (filter = strategies):
  • While some is noise, some others look like driving legit strategies
• artemiusgreat mentioned several sites dedicated to algorithmic trading (I briefly checked the costs)
  • Free: MQL5 Codebase: 
  • 600$-1200$/year:
    • Papers with backtests: 
    • Quantpedia: 
    • QC codebase: 
• The-Goat-Trader (got his comment removed for unknown reasons) mentioned Momentum Rotation and pointed to the sources Gary Antonacci GEM, Ned Davis or Meb Faber 3-Way Model, Andreas Clenow for further informations.

---

If you want to add to the list, just drop a comment and I will edit the post and add it together with an honorary mention of your username. (If two suggest the same strategy twice, time of comment will be the deciding factor).

--

I simply want to implement different strategies and see which is performing which way to test my software and also broaden my knowledge.

Thanks for participating!

I've spent a lot of time researching for the best books to learn algo trading mostly focused on personal use (not to get an algo trading job) and I wanted to share it with you guys in case it would help anyone. With the research I did I tried to organize each category in a logical reading order but of course that is quite subjective.

Its definitely a lot of books and I doubt anyone will read all of them, but maybe it can help you pick a few from each category to learn something new.

If you have any suggestion of books that should definetly be added to the list or removes feel free to let me know! :D

1. Economics in One Lesson (Henry Hazlitt) - 218 pages
2. A Random Walk Down Wall Street (Burton Malkiel) - 480 pages
3. The Little Book of Common Sense Investing (John C. Bogle) - 320 pages
4. Reminiscences of a Stock Operator (Edwin Lefèvre) - 288 pages
5. Flash Boys (Michael Lewis) - 320 pages
6. Trading and Exchanges (Larry Harris) - 656 pages

1. How to Read a Financial Report (John A. Tracy) - 240 pages
2. Financial Statements: A Step-by-Step Guide (Thomas R. Ittelson) - 304 pages
3. One Up on Wall Street (Peter Lynch) - 304 pages
4. The Intelligent Investor (Benjamin Graham) - 640 pages
5. Security Analysis (Benjamin Graham and David Dodd) - 816 pages

1. The Mathematics of Money Management (Ralph Vince) - 400 pages
2. Cycle Analytics for Traders (John F. Ehlers) - 235 pages
3. A Primer for the Mathematics of Financial Engineering (Dan Stefanica) - 284 pages
4. Stochastic Calculus for Finance (Steven Shreve) - 187 pages
5. Time Series Analysis (James D. Hamilton) - 816 pages
6. Analysis of Financial Time Series (Ruey S. Tsay) - 720 pages

1. Python for Finance (Yves Hilpisch) - 586 pages
2. Python for Algorithmic Trading (Yves Hilpisch) - 380 pages
3. Trading Evolved: Anyone Can Build Killer Trading Strategies in Python (Andreas Clenow) - 435 pages
4. The Algorithmic Trading Cookbook (Jason Strimpel) - 300 pages
5. Hands-On AI Trading with Python, QuantConnect, and AWS (Matthew Scarpino) - 416 pages

1. Quantitative Trading: How to Build Your Own Algorithmic Trading Business (Ernest Chan) - 182 pages
2. Building Winning Algorithmic Trading Systems (Kevin J. Davey) - 286 pages
3. Systematic Trading (Robert Carver) - 325 pages
4. Trading Systems and Methods (Perry J. Kaufman) - 1232 pages
5. The Science of Algorithmic Trading and Portfolio Management (Robert Kissell) - 492 pages
6. Algorithmic Trading Methods: Applications Using Advanced Statistics, Optimization, and Machine Learning Techniques (Robert Kissell) - 612 pages
7. Algorithmic Trading and DMA (Barry Johnson) - 574 pages

1. Inside the Black Box: A Simple Guide to Quantitative and High-Frequency Trading (Rishi K. Narang) - 336 pages
2. Algorithmic Trading: Winning Strategies and Their Rationale (Ernest Chan) - 224 pages
3. Stocks on the Move (Andreas F. Clenow) - 288 pages
4. Quantitative Momentum (Wes Gray) - 208 pages
5. Quantitative Value (Wes Gray) - 288 pages
6. The Art and Science of Technical Analysis (Adam Grimes) - 480 pages
7. Finding Alphas: A Quantitative Approach to Building Trading Strategies (Igor Tulchinsky) - 320 pages
8. Active Portfolio Management (Richard C. Grinold and Ronald N. Kahn) - 596 pages

1. Machine Trading: Deploying Computer Algorithms to Conquer the Markets (Ernest P. Chan) - 264 pages
2. Leveraged Trading (Robert Carver) - 346 pages
3. Causal Factor Investing (Marcos López de Prado) - 100 pages

1. Machine Learning for Asset Managers (Marcos López de Prado) - 141 pages
2. Advances in Financial Machine Learning (Marcos López de Prado) - 336 pages
3. Machine Learning for Algorithmic Trading (Stefan Jansen) - 820 pages
4. Machine Learning in Finance: From Theory to Practice (Matthew F. Dixon, Igor Halperin, and Paul Bilokon) - 548 pages

1. Options, Futures, and Other Derivatives (John C. Hull) - 880 pages
2. Option Volatility & Pricing: Advanced Trading Strategies and Techniques (Sheldon Natenberg) - 592 pages
3. Paul Wilmott Introduces Quantitative Finance (Paul Wilmott) - 736 pages

For the past year, I’ve been trying rule-based algo trading, but I haven’t been happy with the results. Now I’m starting to explore machine learning and honestly, it’s confusing as hell. There are so many paths, so many terms, and barely anyone talks about what actually works or which approach is best to commit to.

If you're profitable with algo trading, could you please share:

Which method are you using?
• Rule-based algo trading
• Quantitative strategies
• Machine learning models
• High-frequency trading (HFT)

Also, if you're comfortable, what's your profit factor or average return?

I think this kind of info could really help others (and me) get some clarity. A lot of beginners are stuck trying to figure out which direction is actually worth committing to and the internet is full of overhyped claims with very little transparency.

Would really appreciate honest replies from people doing this seriously.

Edit:
I'm asking this to gain clarity, please don’t misuse the comments to sell your code or claim fake results. Let’s keep it honest and helpful.

Edit: 2

I’ve listed HFT as a separate category, but I understand it can be combined with quant or other strategies. If you're doing that, just mention it.

Basic ability to research and find information on a topic are a must, for one

How would a fairly profitable algo trader start if he were to start today? what all mistakes,loops, could have been avoided? how the start could have been better?

The most important thing is to have a good thesis. If you can’t explain why a strategy works, you are almost certainly trading noise.

The easiest way to think about this is to consider supply and demand. If you want to buy, you should have access to information which suggests there will be more demand than supply for whatever period you wish to trade for. The opposite is true for shorting.

Now ask yourself, how can I determine when one side will have more pressure than the other before it happens? This is why limit order book data is incredibly valuable. In cases where LOB data isn’t available such as in decentralised markets like forex (at least for retail), what else can you do?

That’s usually how I develop my strategies. The method should follow the idea. You shouldn’t start with the goal of using ML to create a strategy. You should have an idea which requires ML to test and verify.

If I could start over, I’d spend 80% of my time on research and risk management, and only 20% on coding.
Most beginners think the edge comes from Python. It doesn’t. Python just executes your idea faster.
I’d learn in this order:
Market structure & statistics
Backtesting pitfalls (look-ahead bias, overfitting, survivorship bias)
Risk management & position sizing
Data engineering (clean data is underrated)
Automation, monitoring, and logging
Biggest mistake I’d avoid? Building a strategy around a backtest instead of building a backtest to validate a hypothesis. The second approach saves months of chasing curve-fitted systems.

Rule-based from a combination of 3-4 indicators / filters (more is too much). Simple things work best for me, just need to find which ones, with the right timeframe and the right asset. (for me it's Nasdaq and S&P500)

It gives me constant returns with 65% / 75% win rate, and a profit factor between 2.5 and 3.5

I was wondering, how ML fits into algorithmic trading? Because of insane amount of noise making it find the edge itself looks like a losing time wasting battle. Are sequential or attention based models any good? Is doing statistical analysis on recent historical data the good way to start. What is the starting approach usually? Do we start with something basic like mean reversion and momentum based approach and try amplify it with ML?

Also whats better? tick level data or different time based candles? I can work around with ML but due to the noise it is just impossible to make any kind of prediction, what are the best practices of keeping the data sane? Thank you for your time.

Rule-based algo. August 10 will mark one year since it started running, and it's currently at 62% net profit

I'm looking for blogs like Turing Finance that write about stuff relative to algo trading.

Any recommendations?

A few more basic ones that come to mind:

Mean Reversion - probably one of the most fundamental concepts, especially pairs trading and bollinger band reversals. From my experience working with trading data, these tend to work better in sideways markets but can get you burned during strong trends.

Momentum - simple price momentum or RSI-based entries. Super basic but surprisingly effective when combined with proper risk management.

Grid Trading - place buy/sell orders at regular intervals above and below current price. Popular in FX markets though it can blow up accounts if not managed properly.

Heyo, I'm a futures algorithm developer running many algos on my accounts with prop firms, and I also do some mentoring on algorithm development. Despite many of the success stories on this forum, most of you are likely unprofitable with discretionary trading. I was trading by hand for over a year before I pivoted to the field of algorithm trading, and has since found some success over the months. I'm hoping to get some people interested, as I personally believe that using algorithms is a more reliable way of becoming consistently profitable.

Currently the community of daytrading in general is incredibly toxic and scammy. That also applies to algorithms, where people are advertising or selling their 3.0 profit factor algos that has practically no drawdown. Beware of those, as they are almost always overfit slop that doesn't perform well live. I will never recommend buying an algorithm from someone on the internet, unless you absolutely know what you are doing or you have good reason to believe they are reliable. But I'd say 99% of what you find out there are garbage. The real way to use algorithms to become profitable, unfortunately, is to make them yourself. Spend some time to learn the statistical methods of algo development, and deploy them. That being said, while many people take a lot of time to develop their own discretionary trading style, it also takes a long time to create reliable algorithms. This is not a silver bullet and there is no shortcut.

As an introductory guide, your main goal is basically two things when creating an algorithm. You want to 1) not overfit your strategy, and 2) have as low of a drawdown (or high sharpe ratio) as possible. These are two introductory ones to get started with, and they are much harder than you think. Don't even consider beating the S&P500 yet. A lot of people don't know how to truly measure that, and if you make something with high sharpe, you're most likely beating it anyway after we apply a concept called the risk of ruin. Just worry about not overfitting while smoothing out your equity curve.

Many people, when making algorithms, just throw conditions and features at historical data indiscriminately and think they've hit the jackpot. Unfortunately, doing so will overfit your strategy and you'll be left with an algorithm that performs terribly live. There is a reason the majority of posts in this subreddit who show off their algorithms only show backtesting results instead of live results. Overfitting is when you capture noise in historical data instead of a real edge, so it looks good in history but fall apart with new live data. It is the number one killer of strategies even for professional algorithm developers; and many people who aren't experienced won't even know that overfitting exists. Instead, there are many methods to reduce the chances of overfitting (you can never bring it to 0%), and if you take the time to learn them, I think it can be very worth it. I personally learned from reading various papers and books that specialize in algorithm development. I've also watched many youtube guide videos that can be good as an introduction to avoiding overfitting. What I'm trying to say is, if you do want to take this route, it will take a bit of time to learn, but there are many good resources out there.

Now for the fun part: a sneakpeak of my own development progress. Currently I have almost 40 algorithms in forward testing, or running live on either prop firm accounts, or paper accounts. I am constantly collecting data and will swap algorithms in and out of my portfolio depending on performance. Here is an example of one being tested, made by one of my students:

I don't normally expect curves to be this good, which is why this one in particular will go through a longer testing phase. Usually I look for a couple of things in the backtesting results, such as sharpe and permutation test p-values, but not as much on the profit factor. But as a general rule I like to tell others, anything above a 1.6 is fishy, and anything above a 2.0 I immediately just dismiss as being overfit or future leaking. Not saying non-overfit algorithms that beat those profit factors don't exist, they're just incredibly unlikely.

Many of my algos are also done with its forward testing phase, after many months and plenty of trades. While I say "done", and by many people's standards a few months is enough data, I personally am not satisfied until I get at least a year. That being said, it's enough for me to hook up to my accounts while monitoring them with care.

This one in particular has too-good-to-be-true live results after over 3 months and 40 trades. I start raising eyebrows when the forward testing results (or out-of-sample) perform better than the in-sample. It's making good money for me right now, but I wouldn't expect performance to continue.

Some edges that my strategies capture are potentially too thin to be reliable. Take this one for example:

High sharpe with low drawdown. Looks good at first glance, but adding slippage can cause the algo to degrade in performance at a greater impact than others. This screenshot has slippage modelled in, but its change after implementation makes me believe the edge is too thin to exploit. Needs to run on simulations for a longer time for me to be satisfied.

Some algos failed somewhat spectacularly in its forward test.

Although, this is technically still within the expectations of drawdown, so I'm not counting it out yet. That's why more than just 3 months of forward testing is required. Is this algorithm overfit or just unlucky? It's hard to tell. That being said, I think it's important to be honest with your failures, especially when it comes to algorithm development.

Anyway, once you develop multiple algorithms, running a few of them in the same portfolio is the easiest way to increase profit consistency. Diversifying and making sure the ones you put into your account don't overlap in strategy type is pretty important. For example, you can run a momentum continuation strategy, with a mean reversion, and also an overnight drifting strategy together. As you develop more, strategy diversity becomes pretty easy and second-nature.

As a final note, algorithm trading is a slow-burner. It's not explosive like some discretionary traders are, and it's not a get-rich-quick scheme. That means losses are common, and red days are inevitable. A good portfolio will rise slowly like bread in oven lol. Here is one of my accounts that run 2 algorithms concurrently, using only 1 micro sizing (last month's performance):

Thanks for reading. I want to reiterate that this is not a silver bullet, and it takes lot of time and effort. I spent way too many sleepless nights coding, sometimes even having nightmares about my algos bagholding losing trades lol. Wasn't fun at the time but is a cool story to tell now.

Are there any good resources for finding some strategies? Is there like a tier list available online or a database?

Hi everyone,

I’m in a bit of a predicament and would really appreciate some advice.

I’ve been using TradeStation to code and backtest my algorithmic trading strategies. I’ve now found a couple that appear quite robust, and I’d like to start forward testing them with paper money.

I attempted to move to cTrader so I could run them in a demo environment, but since I don’t know Python, the process of building cBots has been quite confusing (even with ChatGPT helping).

So I’m trying to figure out the best path forward.

Do you think it’s worth learning Python so I can properly automate strategies in cTrader? Is MetaTrader easier for strategy automation? Or would it make more sense to pivot and run the strategy in a semi-manual way instead?

I’d really appreciate hearing from anyone who has experience with cTrader, MetaTrader, or automating trading strategies in general.

Thanks!

If you already know programming, coding is probably the least important part.

I’d learn market mechanics first. Order types, liquidity, spread, slippage, partial fills, fees, funding, borrow costs, and how your chosen market actually trades. A backtest that ignores execution reality is usually just a good looking spreadsheet.

Next, learn research discipline. Start with a clear hypothesis, not a pile of indicators. Use proper out of sample testing, walk forward validation, regime analysis, and realistic costs. Assume every impressive result is overfit until it survives repeated attempts to break it.

Then focus on data. Point in time accuracy, survivorship bias, corporate actions, timestamp quality, and missing observations matter more than most beginners expect. Clean inputs are part of the strategy.

Only after that would I invest heavily in infrastructure. Use fast vectorized tools to reject ideas quickly, then event driven testing when fills, order state, or portfolio interactions matter.

The biggest mistakes are usually predictable: optimizing across the full sample, trusting Sharpe alone, testing hundreds of variations, building a giant platform before finding one credible hypothesis, and going live before comparing simulated execution with actual fills.

A good starting loop is simple. Pick one liquid market, test one explainable idea with realistic costs, validate it across time and regimes, then trade it at tiny size. Log every difference between the expected and actual result.

What market are you planning to start with?

You will get tired gathering it all together...

• Papers with backtests:
• Quantpedia:
• MQL5 Codebase:
• QC codebase:

It may or may not count as a strategy, but I am finding using multiple timeframes to be highly benificial. Allows for composite analysis of strategies and identification of macro vs micro trends, etc. If your software already allows for granularity in selection of time intervals for strategy development, that would complement most of the strategies you already have on this list, IMO.

I see a ton of posts about reversion and liquidity. Does anyone use algos for more typical strategies? For instance trading within value areas, SPs, filling RTH gaps, a-session POCs, etc?

Yeah, in my opinion some of those strategies just go way too deep. I don’t doubt that they work from time to time; I just like to keep it simple in my own investing. In my case, that means a long-term algo trading strategy with only two assets: VOO and VBIL. By rotating from 100% equities to 100% defensive assets and back based on RSI, moving average crossovers, volatility, and statistical thresholds for rapid declines, I’ve been consistently able to sell VOO at the peak, grow slowly in VBIL while it drops, and buy back in at the bottom for reduced drawdowns and much higher returns. My model very rarely misses rebounds and practically never ends a year in the red. It’s not as flashy as some folks are drawn to, and it certainly won’t make anybody absurdly rich absurdly quick, but it’s returned about 18% annually compared to the S&P 500’s 6% (excluding dividends), with only a few trades each year. I even built in an email loop so now I just get an email a couple times a year telling me to buy or sell. Huge props to those of you who can handle the stress and intensity of complex day trading algos. I just value my sanity and being present with my family more than I value massive returns at this point in my life. 18% CAGR is more than enough for me.

Most of the publically available are least working, may be good for backtesting, but not for real world usage !

You need to find our your own algorithmic edge.

Yeah, many algos actually use simple ideas like trading around value areas, gaps, or VWAP so it's usually nothing super fancy.

I taught myself Python a long time ago. And recently in 2024 I came into Trading. I would love to test my strategies, do trading automatically. Learnt a little bit of MQ5 (for metatrader5) and mostly use Claude for building EA's. They are not perfect, well 99% of the time they don't even work. So you guys could give me tips to start on Algotrading, what routes should I take. In college I don't know what to major in stuck between ComSci, Maths, Finance.
And any free apis like for data over 10 years for backtesting. Backtesting exists already in MT5 But It's not that detailed for me. And also what libraries I should learn in Python,
I know everything I asked here can be found on AI or Youtube or any Blogs online. But I would love to hear from the experienced or people who have been in my position someday.