Self-Exciting Behavior and Detecting the End of Price Trends | Algorithmic Trading Strategy

In this video I show a method I use to exit trades for momentum/trend following strategies. It is the Hawkes Process, a useful tool for algorithmic traders to know. The Hawkes Process is a type of self-exciting process and can be used to handle the self-exciting behavior found in volume/volatility. In this video we apply it to measure of volatility, the range of candles (high - low). Often major trends exhibit higher volatility. Trends often end around the same time that the volatility has dispersed. I show a simple trading strategy using the hawkes process that has decent performance and fairly high win rates for a momentum based trading strategy. A python implementation is shown for the indicator and the strategy.
Patreon: / neurotrader
Code: github.com/neu...
Links
en.wikipedia.o...
en.wikipedia.o...
en.wikipedia.o...
github.com/tr8...
tr8dr.github.i...
tr8dr.github.io/
The content covered on this channel is NOT to be considered as any financial or investment advice. Past results are not necessarily indicative of future results. This content is purely for education/entertainment.

[@Gingeey23]

this must be the best youtube channel for more advanced algo strategies. I appreciate the transparency into parameter impact, and that you don't just show the highly tuned, overfitted results.

[@neurotrader888]

Thanks. Too many times you'll see impressive results only to find it built from 5 suspiciously specific parameters

[@MrRahul15937]

Bro, you are from different planet...

[@arieloxford6790]

Another gem. This stuff is amazing.

[@chechelc]

Man! Your vids are addictive! Keep the great work.

[@sphiwemkhatshwa7980]

I’ve learned so much from this channel - always excited for an upload !

[@RettlonHD]

You need a discord for your community, to discuss such stuff. Awesome video as always!

[@SliverHell]

Neurotrader, do not start a discord for the community, keep things the way they are.

[@MrRahul15937]

Thanks for providing these new set of learning to me...

[@paulparker4078]

Fascinating stuff. Thanks for sharing this, I have learned a lot of interesting techniques from your videos.

[@seanmulherin1055]

Wow, well done. Thanks for including the python code.

[@bongkem2723]

amazing video, thanks bro

[@minimal2224]

I shall develop my own ideation of this now the Hawk-Tuah process.

[@estebangonzalez7036]

One reason why you may not need the mean arrival rate is that you're working with a mean reverting stochastic process by nature which is range or volatility, thus you're already working with a close to 0 mean process. (Disclaimer: I'm not a mathematician, I just deducted that by logic which may be wrong, please correct me if I am!). Great video thanks!

[@neurotrader888]

I don't know enough about point processes/hawkes processes to know for sure (still learning). The mean of the normalized volatility is stationary at 1.

[@OPPACHblu_channel]

thanks for sharing!

[@wootoon]

很有用，期待更多视频。

[@ZVer-gr1ft]

awesome work! thanks

[@cooltraderf]

Really great stuff. I am going to attempt to apply this to options volatility to see if I can find workable signals. Thank you.

[@neurotrader888]

Very cool. Never worked with options myself. good luck!

[@mozkhiyar9486]

Great again as always

[@orjonxhoxhi1314]

Nice work. Keep it up

[@mohammadgh6097]

Great content, keep it up ❤

[@mk6289l]

Which entries work best with these exits?

[@pavanvenkat9182]

Hi , are you a profitable trader using automation?

[@MsRajdutta]

Can you create a video on Donchian channel

[@badredinboudjemaa9668]

Can you pls make a video about how to make the howks process indicator on tradingview

[@NguyenHieu-uw2sx]

Hi do you have any video regarding using AI neutral net to time the market? Thanks,

[@heshammostafafahmy]

Thanks for posting this video but I have one comment about the way the backtest is carried out. Using the signal * returns is not quite a realistic scenario to backtest the strategy. The signals are held to either 1s or -1s consecutively for quite some time. Ignoring the trading fees is actually quite unrealistic as you will end up paying lots of fees and hurts the overall performance of the strategy.

[@neurotrader888]

The idea is a consecutive sequence of 1s denotes a single long position. The fee is only paid on the first 1 (entry) and last 1 (exit). In the period tested the average long trade was 2.9%. A common fee on many crypto exchanges is 0.1%. Of course there is also slippage to consider..

[@heshammostafafahmy]

@@neurotrader888 Right but I suggest you then change the cumsum returns plot as its returns calculated at every candle of the signal without accounting for the trading fees. Realistically this will lead to loses when i tried it.

[@KaiAlgo]

Hey bro, can you guide me what's the best source to learn Quant statistics and math?

[@neurotrader888]

I learned this stuff from reading a lot of books/papers, too many to list. I may make video about books I like at some point. Perry Kaufman's Trading Systems and Methods and the quantstart blog are good resources that I read when I started.

[@KaiAlgo]

@@neurotrader888 what's the keyword when you search about related papers? Would you please explain a little about this?

[@KaiAlgo]

@@neurotrader888 also thanks for guidance. It would be greate if you share the books you read here or in a blog.

[@cheerret]

Я повторил эту стратегию в своей программе с учетом комиссии, и эта стратегия убыточна на большом интервале, как и все другие стратегии из ютуба, увы

[@TopsideXP]

Bro this channel is gonna get taken down by Nancy Pelosi 😭