Recent content by jbax

Also look at the "Chandelier Exit" indicator

You could sell 33% right now, put a tight trailing stop loss on another 33% and a loose stop loss on the remainder.

I run simulations against multiple instruments (40+) at the same time using each set of parameters, collect statistics about the trades made, and from these I select the parameters that worked well enough for the majority. Rinse & repeat for a while, run a few market tests where the simulation...

The "approval" bit is what the OrderManager and StrategyMonitor are for but there is no pre-built risk management routine. You need to provide your own implementation. Backtesting and live trading run concurrently. It was built from the ground up to run concurrently and you can trade live...

That's great, I can help you out. Once you get something done on a public fork I can take a look and lend a hand.

The reason was exactly to prevent memory leaks. That's also exactly the reason why java is used in servers that run for years without stopping. Now, for high frequency trading it's all about preventing the garbage collector to run, so you preallocate everything upfront and only use that. The...

I'm working with java since 2001, almost exclusively. If I look at python code or something else I feel lost too.

Don't forget that pig is part of the backbone of major stock exchanges (NASDAQ, NIKKEI, FTSE, etc). I worked on a distributed system used to route trades from these and the main exchanges of around 50 different countries. It's also used a LOT for high frequency trading (believe it or not, I seen...

Any database with a JDBC driver is supported. Just pass in your sqlite database connection details (jdbc url, user and password) and it will use that without trouble.

Or Scala, or Kotlin.

I use a large combination of factors to determine when to exit. On backtests whenever I use a single method (e.g. Chandelier Exit, Parabolic SAR, Donchian Channels) it doesn't give great results. Stop losses at a fixed price/rate/ATR multiplier also seem to yield worse outcomes if it's not...

I guess his question is more related to the resolution of your data (as in 1 minute candle history for the last 2 years instead of 1 day candle history for the last 2 years). In that case the better the resolution of your data the better your model, and in my opinion the shorter the time frame...

Thanks for the support. In a sense the project is "complete" as in it can be used to build a trading robot for you right now. What needs to be done mainly is integrating with other brokers/exchanges so you can trade live in them. I don't have a lot of incentive to do this myself as I'm not...

That is done on purpose. If you look at the optimizing section, that constructor is changed to use Parameters. This allows you to parameterize not only your strategy but every single thing in your code (bollinger params as you suggested, stop loss percentage, ticks in a trade before getting out...

Some of them yes. quantopian/quantconnect depend on their platform so you have limited control over what you are doing. pylivetrader if I am not mistaken loads all data in memory so it becomes hard to backest efficiently, especially if you are trying to emulate a live market situation (trading...