Portfolio improvements

[DidierRLopes]

• Improve portfolio controller UX
• Shadow commands when these are unavailable
• Allow custom reset when loaded portfolio and benchmark for faster development
• bug when bench was loaded before portfolio, this fixes it
• allow to see sum of portfolio holdings (stacked plot)
• Add R-square command between portfolio returns and benchmark returns for all periods
• Add overall stats command for portfolio and benchmark (mean, std, percentiles, count) for a specific period
• Add kurtosis command for all periods
• Add skewness command for all periods
• Add period parameter to perf command
• Add yearly returns to cumulative return plot
• Add rolling volatility
• Add rolling sharpe
• Add rolling sortino
• Add rolling beta
• Allow to select period for cumulative returns vs EOY returns
• Add volatility, sharpe ratio, sortino ratio and maximum drawdown ratio

@JerBouma just fails on missing docs. Waiting for your approval on this so I can add the docs if you are happy with it.

---

Additions since @JerBouma review:

• remove sqrt(N) and N from sharpe and sortino calculations
• allow hold to export raw data from tail
• automatically add space before and after table
• add portfolio holdings in percentage
• fix relative periods to be more accurate
• refactor metric command to allow to select a metric of interest and see multiple periods for it
• fix cumulative return (in percentage)
• implement new yearly return command
• add daily returns graph
• add distribution of daily returns command and stats for raw
• add monthly returns command
• add summary command with multiple metrics for a specific period
• calculate yearly (out)performance

---

Since 2nd @JerBouma review:

• fix show
• beta with benchmark of 1
• improve meet style

[JerBouma]

Hey, looking good so far! You've been busy this weekend while i was sitting in the sun ;)

Some notes:

• Could you transform the relevant numbers to percentages with 1 or 2 decimals? E.g. stats (since returns), vol, and possibly rsquare.
• Could hold also give me a table (--raw) with the absolute values and % of the portfolio? I know alloc also does the percentages but I think it is useful to see this result in a table regardless.
• return:
  • Could return command also provide me with a table that gives me returns for each year and benchmark and depict outperformance? This combined with the periods I mentioned. Here I do want to point out that I see a clear difference in cumulative returns and the method used within perf (and I would go with perf). The difference here is that the perf method makes an attempt to mimic the scenario that you bought the ETF (benchmark) instead of the asset at every order moment you bought and then determines if that would have led to outperformance. This also makes intuitive sense since I want to understand if I wasn't better off just buying the ETF.
  • I am unsure if the cumulative returns make sense if you are buying on a periodic basis. Why is the cumulative return negative for the portfolio? That makes very little sense looking at how much return you were able to achieve with those early day trades. E.g. look at the return perf gives you.
• I think maxdd and the rolling graphs could all use a table with values.
• rbeta only gives portfolio value but does it make sense we show the beta of 1 for benchmark as well? Just overall consistency and making the graph more clear (??). Optional.
• I'd like stats to also include skewness and kurtosis. I think it makes little sense to have those as separate commands, they have little meaning to look at individually.

Regarding rbeta, see QuantStats:

[jmaslek]

In the case bench isn't supplied, should we just use SPY as default:

2022 May 17, 18:47 (🦋) /portfolio/ $ yret --raw
Please first define the benchmark (via 'bench')

[DidierRLopes]

In the case bench isn't supplied, should we just use SPY as default:

2022 May 17, 18:47 (🦋) /portfolio/ $ yret --raw
Please first define the benchmark (via 'bench')

Good idea, let me add that

[JerBouma]

I disagree with adding a default benchmark. When you use bench you get countless of options already and setting a default benchmark makes little sense if your portfolio is composed of stocks that aren't even in the benchmark. What are you measuring then?

Also, I should go to bed. 🙃

[DidierRLopes]

I disagree with adding a default benchmark. When you use bench you get countless of options already and setting a default benchmark makes little sense if your portfolio is composed of stocks that aren't even in the benchmark. What are you measuring then?

Also, I should go to bed. 🙃

Gotcha, will leave as is then! Will start working on the docs - @jmaslek is helping debug returns