This page acts as an aide memoire for me to keep a record of my ideas for future testing as well as being a list of things I want to do e.g. tweak indicators, compile C++ octave functions etc. As such, this page will be constantly changing as ideas are added, ideas are tested and either accepted and made part of the system or rejected and deleted from the page, the "to do" list items are done. Most of what is on this page probably won't make much sense to readers.
- Oanda_API download script for positionbook
- Create indicator (fmincon?) from positionbook data
- Script to backfill above to match orderbook
- currency strength momentum switching system on 10min bars
- having completed the coding of market_type.cc ensure that all production indicator calculations used as input to the function conform to those used in the MC generation of the market_type.cc look-up tables - specifically
- prices are smoothed in a 1 2 2 1 FIR filter prior to the Cybercycle being calculated
- Tukey chart ucl/lcl calculations are applied to this cybercycle calculation without any further smoothing
- the price_cl_mults & cyber to price cl_mults in production indicators are the same as those used in the above mentioned MC generations
- lookback in production indicators is period + 1
- apply the elliptic smooth in current version of the 1,2,3,4 day oscillator leading signals of cybercyle
- adjust/check coding for adaptive tukey indicator to account for period+1 (done) and reoptimise with MC & do the same for indicators which use full or half period lookbacks i.e. stochastic(done), myrsi(done), adaptive lookback max_min(done), imi(done), lin_predict, highpass(done), bandpass(not nec?)
- apply osc_lead signal with ellip smooth to all bounded oscillators
- redo the zerocrosscount to use the inphase component of the leading oscillator functions (done for inphase_cybercycle)
- all the below in java code for submission to algodeal testing (long term plan)
- slope of control lines as switch for band trading?
- full period repeated median slope as a switch?
- divergences in cyber and fish_cyber
- investigate the crossovers of cyber, smooth cyber and fish_cyber as possible turn indicator
- test these on real prices
- guppy momentum indicator?
- optimise indicator input price (close, typical price, mid price, my price...... ) and the smooth thereof i.e. elliptic, kalman etc.
- full period/other periods of repeated median slope momentum as indicator? divergences? optimise?
- make synthetic price series function period adaptive and incorporate recent volatility aspect
- optimise default repeated median slope using MC synthetic price series in a "run-off"
- detrend via highpass before applying channel bandpass?
- tukey bands around highpass filter trend as indicator
- Bayesian analysis of ols line fit/sinewave fit as market mode switch?
- twicing of the Adam Projection
- apply the reflection principle to set limit entries and stops
- direct comparison of cyber, highpass and bandpass and maybe zerocrosscounts for highpass and bandpass
- set targets based on zerocrosscounts for cyber, highpass and bandpass when market is cycling
- volatility-differentials-highlow-volatility-versus-closeclose-volatility-hlv-ccv/
- the AFIRMA trend line as set up for pyramiding or continuation trades?
- develop a DSP torture test suite
- a cycle quality indicator as in here which looks like this. Use this as a further input to the Naive Bayes Classifier indicator?
- investigate differences in sinewave indicator - see sinewave_3.1.mq4 in Metatrader library
- Ehler's fractal dimension as input to Naive Bayes Classifier?
- Goertzel algorithm to detect period?
- The medcouple adjusted boxplot for extreme moves?
- Adapt the IMI according to this adjusted RSI calculation
- use the VWAP formula according to this link for input price or as a predictive indicator i.e VWAP > pivot is bullish or < pivot is bearish
- use candlestick maths to create new charts
- Structural break package in R
- Andrew Ng's Neural net stuff for market mode
- Enter, and win, a stock picking contest. such as this.
- Investigate time-series-analysis-and-order-prediction to supplement my prediction function.
- Use equity factor analysis, as per this post, for analysis of portfolio level back test results.
- Incorporate the Shark Machine Learning Library into future NN work.
- investigate Hedgeit Python software.
- Time series similarity measures
- investigate thepatternsite.com for ideas
- Ehler's code links.
- Eureqa software
- Runs charts for congestion periods
- Data Smashing
- Turning Point Prediction of Oscillating Time Series using Local Dynamic Regression Models
- Better Hilbert Sine Wave Indicator
- Useful Savitzky-Golay stuff here and here and here.
- Time series features webpage
- Using the generalized lambda distribution to simulate market returns
- Time Frequency Analysis
- power spectral density matlab code
- Time Frequency Representation stuff
- Alpha curves and dtw barycenter averaging
- Effect size pdf and nice webpage
- detecting and measuring lead-lag effect
- Identifying Granger Causality
- Forex NN
- Field Guide to Genetic Programming
- stackexchange questions - how to do exploratory data-analysis for machine-learning
- Link to course from above stackexchange question
- optunity a ML hyper-parameter tuning suite
- Linear Regression
- determinism and entropy - choosing what to trade
- Dynamic time warping and clustering
- Intoduction to ARIMA models
- Independent Component Analysis
- Dr. Keogh's papers
- Extreme learning machines
- Andrew Patton's homepage
- Timothy Masters' homepage
- Causal Impact and its R package.
- datagrapple.com
- deep learning libraries
- measuring market risk
- Quantopian
- QuantConnect
- cloud9trader
- algotrader
- boruta_py all relevant feature selection
- worked example using the boruta package
- select important variables boruta package
- Rattle
- Caret R package and recursive feature elimination
- variable selection with random forests
- R Machine Learning view
- portfolioeffect.com for some useful HF tools
- 15 years of forex tick data to mongodb using python
- mechanicalforex, machine learning, choosing a library
- predicting chaos and r in trading time series forecasting using chaos part 3
- Empirical Dynamic Modelling, the rEDM package for it, the package vignette and Taken's Theorem.
- Bressert Double Stochastic
- automl and tpot for automating the machine learning pipeline.
- financial time series segmentation turning points
- ema trading strategy, full version here, and the Heidke Skill Score
- Reinforcement Learning Github
- residual neural networks
- MLBox, a blog on it, and Entity Embedding.
- currency strength indicator based on log movements of currency indices
- denoising autoencoder (mSDA) based on currency indices
- Yarin Gal - What my deep model doesn't know - application for kalman filter?
- Mitigating over-fitting-on-financial-datasets-with-generative-adversarial-networks and its linked https://github.com/FernandoDeMeer/Mitigating-Overfitting-Experiment
- Pairs trading on my currency strength indices using ideas of kalman-filter-pairs-trading-with-zorro-and-r, the pairs-trading-analysis-with-r course and general pairs trading strategies.
- Weight Agnostic Neural Net training
- DVO (Varadi) Oscillator, applied to currency indices?
