Shared compute is Fincanva computing one result per distinct strategy setup and reusing it wherever that same setup appears, so a strategy whose exact setup has already been computed returns its result immediately. A result belongs to the setup that produced it — instruments, rules, settings, start year, benchmark — not to the account that asked for it.
Also seen as: deduplication, result reuse
How does Fincanva handle it?
- Asking for a setup that is already computing does not start a second backtest of it — you wait on the one under way and get its result.
- Reuse is not a stored badge that can go stale: a reused result is still checked against your settings and the latest market data, so a match that is no longer current is recomputed rather than served (see data freshness).
- Public strategies benefit the most, because many accounts open the same setups — and a separate server-side sweep, not your daily warm, has usually computed them before anyone arrives.
- Nothing about your strategy is shown to anyone else by reuse; what is shared is the computed answer to an identical setup.
What does it look like in practice?
A public strategy has been run many times, so its result exists. You use Copy to Mine and press Run on your copy without editing anything: the setup matches, the finished result is served, and the Compute column reads a near-instant time — no queue, no progress bar. You then change one thing, the rebalance cadence, from every 3 months to every month. That single edit makes the setup new: nothing matches it, the status flips to Needs re-run, and the next backtest computes in full for the first time, with a compute time that reflects the whole work. Change the cadence back to every 3 months and the original result matches again.
Why was my run instant?
Because the answer already existed. Pressing Run on a setup that has been computed before does not recompute it: the finished result is served, compute time reads a fraction of a second, and the status goes to Up to date without ever passing through a visible Computing phase. The reverse also happens: a strategy you have just created can already be Computing the first time you look at it, because an identical setup is already being computed and its result is the one you will get.
What counts as the same setup?
Everything that feeds the backtest. Change one input — an instrument, a rebalance cadence, the benchmark, a cost or tax rate, the start year — and the setup is no longer the same, so no existing result matches it and the next backtest computes in full. Change nothing and it stays a match, which is why duplicating a strategy and running the copy is effectively free. Two strategies identical in every respect are the same question with one answer. The three assumption toggles are the exception that proves the rule: they are not part of the setup, because all of their combinations are computed in the same backtest.