Here's the fastest and easiest way I found to validate a trading strategy:
Monte Carlo Validation
Let's see how a real strategy with an edge compares to a fake system
Summary:
1. Breakdown of a Real and Fake trading system
2. Comparing Equity Curves
3. Monte Carlo Analysis
4.
Noise was my biggest enemy when I started short term trading.
That's why I created anti-aliasing bars.
Here's how I use them to filter out the noise and see the true market direction.
Summary:
- 5-Bit Entropy to measure market noise/signal ratio
- Analysing statistics of 4H
Algo secret sauce
#3893
: How to capture trends with mean reversion.
- Use linear regression with +/- 1.5 standard deviation bands.
- Buy/Sell when close below lower/above upper band and do not have open position.
- When price crosses mid point. Set stop to lock in 50% profit
Here is a breakdown of how I built the algos for my Oil portfolio🧵
Stats:
Profit Ratio: 1.78
Return / Drawdown: 5.15x
Summary:
1. Volatility + Anti-Aliasing Custom Bars
2. Trend Following model = Low Pass Filter + Regime Filter + Stops
3. Mean Reversion = Band Pass Filter +
Algo secret sauce
#8342
: How to avoid whipsaws when trend following.
- When buying/selling the peaks/valleys of a smoothing indicator.
- Place a band +/- 1xATR above and below your entry point.
- Take no more trades whilst the indicator is inside the band.
- This way your
I build all my Algos following a simple mantra:
Simplicity Scales, Complexity Fails.
The trading logic can always be broken down into:
1. Trade Signal
2. Regime Filter
3. Fixed Stops
4. Trailing Stops
5. Profit Locking
6. Drawdown protection
Give it a try!
Think systematic trading is emotionless?
Think again!
It takes nerves of steel to sit on your hands in a drawdown and continue executing.
Here is my Oil portfolio of algos finally breaking out of a multi month drawdown.
A trick I learnt to improve my risk adjusted returns was Regime Filtering.
As my algos are built to capture specific market dynamics, the edge is not always present.
Here are 2 ways I measure profitable periods🧵
@RichardNagy_TFF
I think this was just a really elaborate way to describe a ponzi scheme that failed. Can't pay people at the top if you don't get enough suckers at the bottom.
A huge milestone achieved for my algo platform 🔥
Walk-forward testing is the most important process for trading success, but it's a PAIN to do manually. 😖 Countless hours spent tweaking parameters, saving data... It's enough to make you want to quit.
BUT NOT ANYMORE! My
3. I stopped using HIGH LAG averages. Like SMA, EMA, Hull, DEMA, TEMA etc.
They all had too much lag for the amount of smoothing of the market data.
Instead I started using LOW LAG indicators like the Low Pass Filter and Ehlers Decycler.
I just completed a major step towards launching my no-code algo trading platform:
Applying indicators to indicators.
Check out the UI I've built and let me know what you think! 🧵
Don't be fooled by your Backtest performance.
Your Drawdown can be a lot worse than expected.
As there is no guarantee the markets will behave exactly the same in the future.
That's why I've just dropped this new tool👇
Forget get-rich-quick hype. This is what good algo trading looks like: It's often kinda boring...until it's not. Small losses, sideways movement, then BOOM.
The real game is to stay in the game for as long as possible!
Unpopular Opinion:
Backtests should ONLY EVER be used to check the LOGIC of your strategy.
Everything else (performance, metrics, trades, stress testing etc.) is irrelevant.
A Backtest is not enough to determine if your strategy has any edge at all.
The only things that
Algo secret sauce
#8363
: Always use a stop loss!
- Don't listen to all the posts saying you will make more PnL...
- That's great for making backtests look good... but we are smarter than that!
- Set a stop 4 x ATR(100) distance from your entry point.
- Move the stop to
How to build a 95% win-rate algo.
Please don't scam anyone with this! 🤡
1. Pick any market, flip a coin, heads BUY, tails SELL.
2. Place take profit 5 pips away.
3. Place stop loss 500 pips away.
4. NEVER CLOSE LOSING TRADES.
5. Randomly repeat the above steps.
6. ONLY show
I'm thinking of starting a Prop Firm that will avoid all the gamblers.
The rules will be based on what I experienced trading professionally - Seat Cost and VaR (Value at Risk)
Here are the basic rules:
▶ Can only have 1 account.
▶ No Daily Drawdown.
▶ Max 10% Drawdown of
Algo secret sauce
#1337
: How to re-invest profits without blowing up your account.
- Drawdowns grow larger with time. A 3 year backtest will have larger drawdown than 3 months.
- When trading 1% of Balance you might scale-up right before your biggest drawdown. Don't do that!
1. Fractal Dimension Index (FDI)
Measures the path the price has taken over a period.
If the path is long > 1.4, the market is noisy and choppy.
Avoid taking signals from trend following algos.
Algo secret sauce
#5673
: Position sizing should be based on historical trades.
- The 1% rule is sub-optimal. It is either too small or too big.
- Each strategy should be allocated a portion of your max drawdown - say 2% of a total of 10% DD.
- Then your historical trades (live
Volatility + Anti-Aliasing Custom Bars
The algos use bars that are not based on time, like 1H or 4H etc.
Instead they use custom bars that are built using a combination of volatility and noise filtering techniques.
The benefit of this approach is to print less bars when the
And so it begins... my Funded Trader journey🚀
First day MASSIVE 0.02% on 2 x 0.01 lot trades 🧐
I'll be showing you how I use simple algo strategies executed manually, to steadily grow your account!
None of this buy 5 accounts YOLO 3 of them... might as well play roulette.
𝗥𝗲𝗮𝗹 𝘁𝗮𝗹𝗸: Your trading journey should be about progress, not perfection.
Every trade, every strategy you build, teaches you something.
Embrace the lessons and keep refining. 🤖
After daily follower cleaning of cam girls and crypto casino giveaway bots.
I finally crossed 400 followers... again.
Thank you, all non-bot followers. So much cool shit in the pipeline to share with you, it will blow your mind 🤯
1. I stopped using Crossovers as entry signals.
Crossovers have too much lag, and I always entered the new Trend very late, leaving profits on the table.
Instead I started using the peaks and valleys of the moving average as entry signals.
I got into the trends much earlier
@bensparham_
yea its like constantly coming back to the Roulette table after losing all your money time and time again, eventually winning and then bragging about it haha
I build algos following these 3 simple principles:
1. Identify repeatable patterns.
2. Develop simple, rule-based strategies.
3. Thoroughly backtest before going live.
It is simple but not easy.
Trading from the chart and just your gut feel is completely nuts.
I've been trading the prop firm account as a discretionary trader to remind myself of what its like.
How can you not second guess yourself without data to backup your decisions!?
Might as well flip a coin...
Imagine hundreds of tireless algorithms, scanning every market, every timeframe, hunting for the perfect trade. This is the power of my algo portfolio.
Humans don't stand a chance!
#AlgoTrading
My systematic strategies are not about catching every market move.
They are about consistency.
Focus on making small, repeatable gains. In the long run, they stack up.
Earning big with algo trading? It's possible, I've done it! 💰
Below are some golden rules:
- Let repeatable logic drive your decisions.
- Build simple strategies for specific market conditions.
- Test rigorously, so you have confidence pulling the trigger every time!
When I was a pro trader, I would make money and buy stuff.
I thought this is what happiness looks like.
When I moved from the UK to Portugal, I threw away/donated/sold 95% of everything.
It was the greatest feeling in the world.
Don't let your possessions own you.
Check out the demo of my game-changing update: one-click walk-forward analysis 🤯
This isn't just an incremental improvement; it's a paradigm shift in how we develop and validate algo strategies.
Efficiency: No more wasting countless hours on manual testing. Run multiple cycles
@EAForexAcademy
Judging by how your EA operated: grid trading with probably martingale position sizing; its pretty obvious to spot and thats probably why you got banned.
2. Market Meanness Index
Measures the likelihood of the market returning to it's Median value.
If the average MMI is falling, this could be an early sign of a trend forming.
The most important advice for algo trader. Never pick the best parameter value after optimisation. Instead pick the middle of a stable, broad plateau.
#algotrading
Robustness test: Stability over a wide range of parameters
Let's look at a specific example:
A trendfollowing strategy that trades a breakout of a 30-day high. We did not optimize the period - i.e., we set the 30-day high because we want to trade stronger longer-term breakouts.
@pedma7
Quantopian tried this same approach for 9 years and eventually shut down in 2020. I think that lot's of people are able to produce good looking backtests, but its a different world to actually consistently generate profits long term.
Traders are slowly finding the holy grail - a diversified portfolio of simple algo strategies. More robust, better risk adjusted returns. Easier said than done tho as I spent 3 years building my own tech to make this happen!
Fantastic post on the holy-grail of trading - Portfolio Trading. The key to building one is using simple uncorrelated strategies across multiple markets and timeframes. Or just let Spectral Alpha do it for you!
What are the main benefits of portfolio trading?
Better capital allocation
If you trade uncorrelated strategies you can use the same capital and share it between them.
Better account performance thanks to higher capital usage
With better capital allocation, your trading account
2. I started using longer periods to capture the major trends on lower timeframes.
Using periods of 500+ on the 1 hour timeframe stopped me from getting whipsawed in choppy markets.
Challenge for the day: Build a way to construct an algo without any code.
What are the best no-code strategy creator tools you have seen/used?
I'll go first...
I was always scared to adjust my system parameters after it went live.
I realised that markets evolve, and my strategy should too.
Retraining parameters kept my strategy in tune with the markets.
@ZenomTrader
mate, over the next few weeks ill be launching my no-code algo trading platform. I think it would save you a bunch of time and hassle with developing and validating your algos!
@propsafeio
@SamBradburyFX
Sam you are wrong. If I have 100s of algos that trade 0.01 lot sizes. Does that mean I'm cheating? No of course not, im managing risk and diversifying. Banning someone because they trade 0.01 is nuts.
𝗡𝗼𝘃𝗶𝗰𝗲 𝗧𝗿𝗮𝗱𝗲𝗿: I'm chasing the latest trending strategy, hoping to get rich quick.
𝗘𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲𝗱 𝗧𝗿𝗮𝗱𝗲𝗿: I'm focused on disciplined execution of my proven strategy, aiming for sustainable long-term gains.
My life 4 Years Ago:
- Long commute
- 12 Hours in the office
- No time for hobbies or exercise
Now:
- Living on the beach
- 4 hours of focused work
- Making music, swimming in the ocean
Take risks, your future self will thank you.
@PJSmith
This is completely nuts. Of course if they were a real prop firm they wouldn't care how you made money for them. But because their business profits come from failed challenges/fees they don't actually want you to succeed...
This is what it's all about for me! 🥰
Sharing my hard earned experience on how to build better systematic strategies.
So you can put them into practice and transform your trading.
Thank you!
4/4 Less lag = better trading
It is now much easier to distinguish between choppy and trending periods.
There are few bars that cover the same span of time, with clearer direction.
So Trend Following algos can use shorter averaging periods => less signal lag.
Also, with less
My tribe has grown 100% in 3 weeks.
400 -> 800 followers.
A massive thank you to each and every one of you.
At 1000, I'll do a massive giveaway.
Can we get there!? 🚀
It's a Monte Carlo Engine that uses your Equity or Balance data to calculate the Expected and Worst Case scenarios.
Use it for free here 👇
It works by sampling your data without replacement.
This keeps the overall gain and produces many other equity
1/4 How to measure market noise?
I found shorter timeframes difficult to trade because of the small fluctuations generated by all the different market participants.
The markets constantly drift between directional moves and choppy action.
This can be measured using signal
Check out the demo of my game-changing update: one-click walk-forward analysis 🤯
This isn't just an incremental improvement; it's a paradigm shift in how we develop and validate algo strategies.
Efficiency: No more wasting countless hours on manual testing. Run multiple cycles
300+ followers strong & growing! 🎉 Thank you for being part of my algo journey. To celebrate, I will be sharing some BIG updates:
🔥Revamped platform based on YOUR feedback,
🔥Live trading sessions,
🔥Funding challenge wins.
@bensparham_
@DarwinexZero
mate there is literally no excuse for not having a DarwinexZero account nowadays.
Obviously apart from not being able to trade 😆
This fantastic post just confirmed why it's taken me YEARS to build my algo trading platform 😅
Goal: Make it so easy to run algos, so you won't even need to read massive guides like this!
There are seven easy steps to follow when creating a
#trading
system. If you're starting out in this field, take a look at the common structure that any algorithmic system should have:
A. Data Ingestion and processing:
-Connect to market data sources or APIs
-Receive and parse
There are many more steps needed to train, validate and deploy algos for live trading.
I share how to do this and more with my newsletter subscribers.
Absolutely free to sign up here 👇
Trend Following model = Low Pass Filter + Regime Filter + Stops
The Low Pass Filter (LP) removes the high frequency noise from the market data.
Trade signals are created when the LP is rising or falling.
A regime filter measures how likely the market is to mean revert. Signals
I took me around 56,979 hours to become a professional trader at one of the world's largest commodity trading houses and an additional 26,298 hours to develop my own algo trading tech.
You must be patient to succeed.
2/4 How much noise does 4H Oil market have?
1 Year of 4H Brent Crude Oil bars have the following stats:
Entropy: 4.98 => Very Noisy
Average size: 36 pips
@UmarAshraf
If you waited for a 100% conviction in any trade, you would never execute anything! The future is unknown and you need to be prepared for all eventualities!
@MeetMattL
Even more evidence of the ponzi scheme.
If they were hedging your trades after you got funded, they would not care how much money you make and how. As they would also be making money. Hence the "profit split". Of course, the money only comes from failed challenges.
@sam_kb_
Mate, if all these twitter traders got together, they would form the best, most powerful hedge fund in the world.
Ray Dalio has nothing on these guys...
Time based charts are cool and all... but why should your trend following algo get whipsawed cos the market is out at lunch? 🤔
Coming soon... non time-based charts that compress periods of inactivity but still react to events.
Dripping with alpha 💦
I got through more than 20 business books.
I applied exactly ZERO principles when starting my own.
Here are the 2 most important lessons I learnt the hard way:
— 1. Build an audience early —
When you get the idea, speak to as many people as possible.
Don’t be scared that
Again more confirmation that an automated portfolio of simple algos is the HOLY GRAIL of trading. Easy to explain... very difficult to achieve.
Ditch all these chart based gurus and their "secret" strategies. Diversify. Automate. Win.
@MeetMattL
@Tonycapalot9
I think you should stop misleading traders by calling yourself a "prop" firm. Evaluation, challenge, demo, practice sure, your business model isn't to get returns on your company's capital, its just from failed challenges.
Mean Reversion = Band Pass Filter + Regime Filter
The Band Pass Filter is great for finding the peaks and valleys of mean reverting markets. By only focusing on a certain frequency range.
The trade signals are generated to buy/sell before the market turns around.
A regime
@quant_arb
In my experience, I can't count the number of times my quant team brought me a tearsheet that had an epic 6-10x return on DD every year for the last 10 years, because they ran a bazillion parameter optimisation.
You can eyeball it, only when you have skin in the game
When I was trading based on the news it felt like riding a bike.
I got better with time and experience.
Now my automated trading feels like driving a Formula 1 car.
Exhilarating with a risk of crashing and burning.
A super trick if you are chart based trading: Fixing the channel to a window.
- TradingView has a drawing tool for the linear regression channels.
- Select a START and END point where you have identified a rangebound market. The channels are automatically drawn:
- Make sure to
For all chart based traders that draw support/resistance, liquidity, fair value or other types of lines.
Which ONE of the charts below is the real market?
Comment what you think!
When I first started, I thought trading had a secret formula.
Years later I learnt it's a journey of consistent learning and adaptation.
𝗠𝗮𝗿𝗸𝗲𝘁𝘀 𝗮𝗿𝗲 𝗮𝗹𝘄𝗮𝘆𝘀 𝗲𝘃𝗼𝗹𝘃𝗶𝗻𝗴, 𝗮𝗻𝗱 𝘀𝗼 𝘀𝗵𝗼𝘂𝗹𝗱 𝘆𝗼𝘂.
Something I learnt the hard way:
Systematic trading is not about timing the market.
It's about time in the market.
𝗬𝗼𝘂 𝘄𝗮𝗻𝘁 𝘁𝗼 𝘀𝘁𝗶𝗹𝗹 𝗯𝗲 𝘁𝗿𝗮𝗱𝗶𝗻𝗴 𝟭𝟬 𝘆𝗲𝗮𝗿𝘀 𝗳𝗿𝗼𝗺 𝗻𝗼𝘄.
@paladinbhutia
I've had weeks where I felt like I couldn't win a coin flip. 😂 But trading small means those losses are just speed bumps on the road to the bigger picture.
@choffstein
Best point is not letting the computer dictate what you should be doing (i.e. data mining bias)
Instead you should be taking what you know works and turning it into a system.