Loc Nguyen’s Reviews > Systematic Trading: A unique new method for designing trading and investing systems > Status Update

6. Instruments: not yet going into the framework, just an overview of how to decide on instruments
a) Min req: open data, accessible min size, controllable price drivers, some volatility; b) instr fit w style: max no of instrs; low correlation for diversification; low cost & high liq (high cost & low liq = slow trade); balance skew of instr w style (e.g. use (+) skew rule on a (-) skew asset); c) access methods

6. Instruments: not yet going into the framework, just an overview of how to decide on instruments
a) Min req: open data, accessible min size, controllable price drivers, some volatility; b) instr fit w style: max no of instrs; low correlation for diversification; low cost & high liq (high cost & low liq = slow trade); balance skew of instr w style (e.g. use (+) skew rule on a (-) skew asset); c) access methods

4. Portfolio Allocation
a) Naturally, higher weights on high SR and low corrs; b) Bootstrapping to avoid over-optim: basically averaging results across multiple optims (instead of one optim on the average) - see Appendix C; c) Handcrafting portfolio wgts: basically norming across risk, corr, industry etc - see clustering methods; d) Augment with Sharpe, not through returns but costs.
A chapter that feels strange.

3. Fitting
Overall: a) Across industries, mixture of experts perform well; b) There's a balance between generalizability & specialized edge;
Tactically: a) KISS; b) out-sample testing only; c) don't drop (highly correlated) rules without testing carefully; d) pool time series data if possible; e) don't compare neg vs pos skewed strat; f) remember benchmarks
Author recs starting with 1. trend following & 2. carry.

2. Systematic Trading Rules

a) sourced data-first or ideas-first - ideas-first is preferred but may need more intuition (?)
b) there's a gap in quantifying data for "merger arbitrage" (?)
c) skew needs to fit with trading style
d) providing liquidity is different from holding illiquid assets
e) asset allocator tends to risk-parity investing
f) crowded trades are deadly (but how to detect?)

Chapter One on psychology is not purely just fluff but also discusses volatility & how to calculate it on a back-of-the-envelope basis.

Pausing at p28. Author proposes that trading strats/rules should be modular -- not sure yet how he'll accomplish that.

The classification system of players into (a) allocating investors (b) semi-automatic traders (c) staunch systems traders -- then building example strategies (called "rules" by the author) is a good pattern for a book.

Steps: 1. classify your audience; 2. convince them separately of your value; 3. give concrete examples for each audience