Allocation is how a strategy divides the capital available to it among the things it holds, and the allocation method is the rule that decides each share. The method turns a list of candidates into a list of weights — one percentage per item — and those weights are what the backtest actually buys. Fincanva runs the same machinery at two levels: a Combined allocates across the strategies it holds, and each strategy allocates across its own instruments.
Also seen as: weighting, position sizing, capital split
Why does allocation happen at two levels?
Because a Combined and a strategy are allocating across different things, so each needs its own rule. The Combined level decides how much of the total capital each member strategy receives; the strategy level then decides how that strategy's slice is split across its own instruments. Which instruments those are comes from the strategy's universe — a hand-picked basket, or whatever a screener returns from its seed universe.
The two levels are independent layers: Fincanva applies both, one after the other, and never flattens them into a single list of instrument weights. You read a Combined's split across its strategies and each strategy's internal split as two separate answers. A standalone strategy only ever uses the strategy level.
Which allocation methods can I choose?
Eleven methods are available to a strategy; six of those eleven are also available to a Combined.
| Method (as the app labels it) | Across strategies in a Combined | Across instruments in a strategy |
|---|---|---|
| Equal Weights | yes | yes |
| Fixed Allocation | yes | yes |
| Inverse Volatility | yes | yes |
| Ranking-Based | yes | yes |
| Risk Parity | yes | yes |
| MPT (Markowitz) | yes | yes |
| Market Cap | no | yes |
| Min Correlation | no | yes |
| Beta Neutral | no | yes |
| Mimicking | no | yes |
| Floating | no | yes |
In prose: Equal Weights, Fixed Allocation, Inverse Volatility, Ranking-Based, Risk Parity, and MPT (Markowitz) work at both levels. Market Cap, Min Correlation, Beta Neutral, Mimicking, and Floating work only inside a strategy, across its instruments — they are not offered when a Combined splits capital across its member strategies.
Each method page covers its own rule; which of them read history and which do not is set out in Calculation window. How a Combined's split across its member strategies differs from a strategy's split across instruments is covered in Combined weighting.
Does a method I am not using still affect my results?
No — only the method you have selected executes, and every other method's settings are inert. Each level holds one selected method at a time, and switching to a different method resets the previous method's own parameters: the picker warns "The current method’s specific parameters will be reset." before it applies the change. Nothing a non-selected method was configured with reaches the backtest.
What do all allocation methods have in common?
Every method, at either level, shares the same three pieces of vocabulary.
- Weights. A method's output is one weight per item, as a share of the capital being allocated. Weights are relative: you can enter raw numbers that do not add up to 100, and they are standardized so the allocated capital is fully used. A negative weight means a short position, which only the methods that support it can produce — see Direction.
- Rebalance interaction. The method recomputes its weights at each rebalance date, and the strategy trades back to them. Between rebalance dates the weights drift with prices. Floating is the deliberate exception: it lets weights drift and realigns them on its own schedule.
- Calculation window. Methods that read history — volatility, correlation, beta, ranking — read it over a window the app calls In-sample, in months. Its hint reads: "Historical window used by the active method for volatility, correlation, beta, and similar calculations. Default 12." Equal Weights, Fixed Allocation, Market Cap, and Floating do not use it — see Calculation window.
Separately from the method, each level carries one dial for how much capital is put to work: Leverage inside a strategy, and the invested portion at the Combined level — see Invested capital and the cash reserve.
Defaults in Fincanva
- Equal Weights is the starting method at both levels, so a new strategy or Combined splits capital evenly until you change it.
- The In-sample calculation window defaults to 12 months, and Leverage defaults to 1.00 — no leverage.
- Each level holds an allocation profile per risk regime: Risk-On, used while no risk condition is firing, and Risk-Off, used while one is. A strategy with no risk condition configured has only the one profile, and the app says so: "No Risk conditions configured — this allocation runs at all times."
- Risk-On and Risk-Off each pick their own method, so a strategy can allocate one way normally and another way defensively.
- With a single instrument there is nothing to divide: "With one instrument and no Risk-Off split, 100% of capital goes to that instrument."
Worked example
A strategy holds four instruments — A, B, C, and D. Under Equal Weights the method returns 25% each, and the backtest buys a quarter of the strategy's capital in each. You switch the same strategy to Fixed Allocation and enter 40, 30, 20, 10; the method now returns 40% / 30% / 20% / 10%, so A gets four times what D gets. Same instruments, same dates, same rebalance cadence — only the allocation method changed, and the results differ because the capital was split differently.
Now put that strategy into a Combined alongside two others and leave the Combined on Equal Weights. The Combined gives each of its three member strategies one third of the total capital; inside its third, our strategy still splits 40 / 30 / 20 / 10. Instrument A therefore ends up with 40% of one third — about 13.3% of the Combined's capital. That is the two levels, applied in order.
Where this term is used
Generated · 3 pagesThe pages that reference this term — so a term page is somewhere you pass through, not somewhere you land and stop.
Fincanva provides no financial advice. Backtests show what would have happened — not what will.
GLOSSARY · 193 TERMS