Recently, I've been frequently checking plaza content out of boredom and found many so-called 'quantitative model' masters who always make money.
This article is a popular science text mainly helping everyone understand what true quantitative trading is to prevent being scammed.
1. What is a true quantitative trading model?
Quantitative trading essentially refers to the use of mathematical statistical models, historical data analysis, and programming techniques to build trading strategies based on objective data, aiming to reduce subjective judgment and emotional interference, and enhance trading efficiency and robustness.
Core features:
Clear logic: Each trade has a clear signal source (e.g., moving average crossover, Bollinger band breakout, etc.).
Backtestable: Strategies can be evaluated using historical data.
Risk controllable: There are clear stop-loss, position management, and capital allocation mechanisms.
Reasonable returns: The win rate may not be high, but stable returns can be obtained through long-term compounding.
True quantitative trading is a technical activity that integrates mathematics, finance, and computer science; there are no guarantees of 'sure profits' or 'get-rich-quick' schemes.
Key point: Quantitative models are not metaphysics; trading is also based on certain indicators or a composite of certain indicators, with an overall win rate generally between 50%-60%, and profit-loss data will fluctuate and rise within a reasonable range (with risk of loss).
2. What are the scams in quantitative trading?
Example:
High win rate and stable profits
The content in the two images is obviously not a quantitative model, but images stolen by scammers or trading history records built through pairing large funds with small positions using a martingale strategy (Sharpe ratio is extremely low, even worse than keeping money in the bank for interest).
2. Disguise as professional using 'pseudo backtesting' or 'fake profit charts'
Slogan: 'This is our profit curve over the past 12 months, steadily rising, with no drawdowns!'
✅ Reality:
Backtest results can be easily faked, such as training models with future data (data leakage issues), only showing profitable accounts, or directly photoshopping charts.
Model parameters overfit history and are extremely unstable in reality.
PS: With this method, I can write a backtested quantitative model with a win rate of over 98 in five minutes.
3. Pretend to 'freely' configure quantitative models
Slogan: 'Free to use, helps remotely configure, no operation needed'
✅ Reality:
Complete API and wallet authorization by remotely changing the exchange language (equivalent to account theft).
If you don't understand, feel free to leave a comment, and I will answer one by one.