Activity
Mon
Wed
Fri
Sun
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
What is this?
Less
More
Owned by Tim
Master AI Trading with QuantConnect and Claude Code. Build and run your own algorithmic trading system, even if you’re not a developer
13 contributions to QuantConnect Mastery
1d •
Case Study: Semiconductor Momentum Strategy (101.9%% CAGR, 1.830 Sharpe)
Three concentrated semiconductor holdings with three hedge layers returned 101.9% a year at a 1.830 Sharpe ratio, which is what the best-looking numbers in this library look like. Over thirteen held-out years the same engine returns -16.70% at a 94.2% drawdown, and adding the hedges makes it worse. It is a bet on one era, and it can be held as one knowingly. The full write-up, including how it was built and what failed, is on the case study page
1 like • 7h
@Elizabeth Rose thank you. 😊 Its important to understand all sides of your strategy and negative sides tells more than positive. Most people they want to see what they want, not what data tells them.
0 likes • 2h
@Elizabeth Rose thanks. Your opinion is very important for me. I'll share everything I have, without hiding anything.
6d •
Which finantial instruments do you want to trade?
Choose the one or drop in a comment ;)
Poll
4 members have voted
0 likes • 6d
As for me its futures, stocks and probably some options.
0 likes • 9h
@John Schuberth great to have an option trader among us. I see you've been trading options for quite a bit. What challenges do you currently face?
9d •
Case Study: Momentum Pullback Strategy (13.8% CAGR, -20,7% DD)
Buying two-day flushes in strong large-cap uptrends returned 13.8% a year at a 20.7% drawdown against the S&P 500's 10.9% at 55.0%, and it made 4.3% through the financial crisis while the S&P 500 lost 54.1%. The full write-up, including how it was built and what failed, is on the case study page
0 likes • 4d
@Zoran Jovanovic How about the contest, lets choose a community QC strategy and try to improve its performance and then compare results. What do you think?
0 likes • 4d
@Zoran Jovanovic cool. Choose any strategy you like there and send it to me. Don’t forget to measure the starting metric as soon as you clone the strategy.
4d •
🚀 New Video: QuantConnect Backtesting Tutorial: How to Backtest Any Strategy with Claude Code
In this video I walk through the whole process from scratch, step by step. I open a QuantConnect account and a Claude account, connect Claude to QuantConnect through the MCP, open Claude Code, and then use it to backtest a strategy on QuantConnect. At the end I go through the backtest results so you know how to read them and what the numbers actually mean. ⚠️ By the way, to use QuantConnect MCP you need to pay for Assistant Compute Node on QuantConnect whereas if you use Strategy Research Lab (which is provided to members of this community) then you don't have to pay for it and still run all backtests. Thats 24$/month savings.
0 likes • 4d
@Zoran Jovanovic I believe so too, because MCP returns much more data then targeted toolkit. This returned data is stored in a context and consume tokens. So it make sense.
20d •
Backtesting node
Hello everybody, Just wanted to know what node subscription everybody has for their backtesting with QuantConnect. I'm currently subscribed to two nodes on the. B2-8. Anyone else have something different that is generating better performance, or is everybody relying on the free node?
0 likes • 5d
Guys if you have any feedbacks don’t be shy to share them ;)
0 likes • 5d
@Zoran Jovanovic from my end, I can say that it was always also happening until I systematized all the processes. What was happening before: I wrote all the toolkits for myself to create the strategies, process them, sweep them, and pretty much everything which Strategy Research Lab provides. What Claude was doing was: 1. Getting all this data and analyzing it. 2. Coming up with a script which will run it. 3. Coming up with a different script which will get the results. 4. Coming up with a different script which would analyze it. 5. Coming up with a different script which would do something else. Therefore, it was actually pivoting from the point of research it must be focusing on to all the other stuff. That was actually the reason why I systematized everything and put everything into one toolkit. Right now, Claude just knows that, in order to create a strategy, he needs to use that. In order to test the strategy, he needs to use that. In order to get the result from the strategy, he needs to use that, and he actually uses the right toolkit when he needs to. It doesn't spend any more time doing the wrong thing. It only spends time actually researching the strategy and thinking about how to improve it.
1-10 of 13
@tim-sergeev
Helping people build profitable algorithmic trading systems.
Active 1h ago
Joined Jan 8, 2025
Citizen of the world
Powered by