Every euro, on the day you added it
A strategy that beats buy-and-hold is only interesting if the buy-and-hold number was honest in the first place.
Aurono shows you two comparisons on a strategy page: what the same money would have done if you had simply bought the coin and held it, and what the same money would have done drip-fed in on a fixed monthly schedule. Both answer a question worth asking. Until v0.1.34, both were computed on an assumption that quietly flattered or punished a strategy, depending on how you had funded it.
This post is about fixing that assumption. It works through two example strategies, down to the individual trade. Both are based on real strategies running with real money, with the numbers rounded so the arithmetic is easy to follow.
The assumption
The old comparison took the net amount you had put into a strategy, and pretended all of it had been there on the first day.
If you funded a strategy once when you created it, that is exactly right, and nothing in this post changes your numbers.
If you added money later, the assumption is wrong in a specific direction: the later money was credited with buying at the price on the day the strategy started, not the price on the day you actually paid it in. And if you had ever taken money out, the error was worse, because the withdrawal was subtracted from the whole history, including the months before you took it out. The comparison held a smaller position than you had held, for a period when you had not yet withdrawn anything.
The fix needs each movement’s date rather than a single total, which is why it was not there from the start: every euro is now put to work on the day it arrived, and a withdrawal sells out of the position on the day it was taken.
A worked example the change does not touch
To see what the comparison actually measures, it helps to start with a strategy where the old and new behaviour agree exactly, so the arithmetic can be checked without the change getting in the way.
Example 1: a BONK strategy funded once. One deposit of €30 on the day it was created, early in September, and nothing added or removed since. The strategy checks the price every four hours, buys €10 when BONK drops about 4%, and sells €10 when BONK rises about 4%. Because there is only one movement of money, “all of it on day one” and “each euro on its own day” describe the same thing, and both versions of Aurono draw an identical line.
BONK trades at a fraction of a cent, so the prices below are per million BONK. Over the three weeks of the example, BONK went from €2.89 to €3.27 per million, a rise of 13.1%.
Here is every trade the strategy made:
| Date | Side | Units | Price per million | € |
|---|---|---|---|---|
| Week 1 | buy | 3.70 million | €2.70 | 10.00 |
| Week 1 | buy | 4.35 million | €2.30 | 10.00 |
| Week 3 | sell | 3.57 million | €2.80 | 10.00 |
| Week 3 | sell | 3.45 million | €2.90 | 10.00 |
Two buys, two sells, and two completed round trips returning €1.13 and €1.43. Total realized: €2.56.
The strategy still holds 1.03 million BONK, worth €3.37, next to €30.00 in cash. Its value: €33.37.
Three answers to the same three weeks
| Buy and hold | Monthly DCA | This strategy | |
|---|---|---|---|
| Entries | 1, at €2.89 | 1, at €2.89 | 2, averaging €2.48 |
| Exits | none | none | 2, averaging €2.85 |
| Holds at the end | 10.38 million BONK | 10.38 million BONK | 1.03 million BONK and €30.00 cash |
| Value | €33.94 | €33.94 | €33.37 |
| Return on €30 | +13.1% | +13.1% | +11.2% |
The strategy page sums this up in one sentence: -€0.57 vs buy & hold and -€0.57 vs monthly DCA.
The two benchmark columns are identical, and that is correct rather than a bug. Monthly DCA buys once per calendar month, and a three-week window inside September contains one month boundary: the first day. With one scheduled purchase, drip-feeding and buying everything at once are the same thing. On a strategy that has run for several months the two lines separate.
Why it trailed buy-and-hold
Two effects, pulling in opposite directions.
The strategy bought 14% cheaper than buy-and-hold did. Buy-and-hold has exactly one entry, on whatever day you happened to start: €2.89. The strategy waited for its drop trigger twice and averaged €2.48. The second buy at €2.30 is where most of that difference came from, 20% below buy-and-hold’s entry. Buy-and-hold could not add there. Buy-and-hold had spent everything on day one.
Against that, two things cost more than the cheap entries earned. About €10 of the €30 never left cash, because the strategy bought only twice. And the strategy sold 87% of the coins it bought at an average of €2.85, 12.9% below the final price, before the last leg up.
That is the whole story, and it is 57 cents on €30.
What the strategy gave up
The strategy is now mostly in cash. From here, if BONK keeps rising, buy-and-hold pulls further ahead and the strategy can only follow with its remaining 1.03 million BONK. If BONK falls, the strategy keeps most of its €33.37 while buy-and-hold gives the gain back.
Being mostly out of the market is a position, not a neutral state. The comparison does not reward or punish that position. The comparison reports it.
Where the change actually bites
None of the above moved by a cent under the new comparison, because the BONK strategy was funded once. The second example was not.
Example 2: a HYPE strategy with withdrawals. It started in May with €100. Early in June €50 was taken out, in July another €11 to fund a different strategy, and in August €20. That is €100 in and €81 out, a net of €19.
The old comparison took that €19, pretended all of it had been there on the first day, bought 0.38 HYPE at €50, and valued them at the final price of €76: a buy-and-hold figure of €28.88.
Dated properly, the same money buys and sells at the prices that were actually on the screen when it moved:
| Date | Movement | HYPE price | Effect on the comparison |
|---|---|---|---|
| May | +€100 | €50.00 | buys 2.00 |
| June | -€50 | €62.50 | sells 0.80 |
| July | -€11 | €55.00 | sells 0.20 |
| August | -€20 | €80.00 | sells 0.25 |
That leaves 0.75 HYPE, worth €57.00.
Every withdrawal happened at a higher price than the €50 the €100 went in at, so each euro taken out cost fewer coins than the old comparison assumed. The honest buy-and-hold kept almost twice the coins, in a coin that rose 52%.
For the strategy, this correction is bad news. The strategy ends at €31.60: €24.00 in cash and 0.10 HYPE, after six round trips that realized about €8. Under the old comparison the strategy page said +€2.72 vs buy & hold: slightly ahead. It now says -€25.40 vs buy & hold, and the comparison with monthly DCA moved from ahead to behind as well. The strategy spent most of a rising summer in cash, and the old comparison could not see that, because the old comparison had removed the withdrawals from a period when the money was still in.
How often the dates matter
You might expect most strategies to be funded once and left alone. Once you move money between strategies with the Rebalancer, that stops being true: most strategies end up with a history of deposits and withdrawals, and for most of them the buy-and-hold figure changes. The change can go against the strategy, like the HYPE example above, or in its favour.
On a real Zcash strategy the correction went the other way. About €43 was added after the start, when ZEC cost between 1.4 and 4.1 times its starting price, and the old comparison had credited that money with the starting price. The buy-and-hold figure of that strategy dropped by about €43.
The direction depends entirely on where the price sat when your money moved. That is the point: this is a timing correction, not a thumb on the scale.
What this does not fix
The comparison is now honest about when your money was in. The comparison still reports the result against your net contributions, and when money has moved both in and out, that net number can be much smaller than the amount that was ever actually at work. The HYPE example shows its figures against a €19 base, because €100 went in over its life and €81 came back out.
The euro figures are right. A percentage computed from them measures against a base that shrank when capital was returned, which answers a different question again. That question needs a different measure, and we are working on it separately.
Where you will see this
On a strategy’s page, in the sentence at the top and in the Benchmark card lower down. Both read from the same series, so they agree. The same release also stopped measuring coins you already owned at your old average cost: a strategy that started with them is now compared from what those coins were worth on the day it began.
A strategy that has not completed a round trip yet still shows its comparison, and says so. It was tempting to hide the number there, on the logic that a strategy which has never sold is just holding. That logic is wrong: a strategy whose triggers have not fired is sitting on your cash rather than being fully invested, and that difference is exactly what the comparison measures. Hiding it would have removed the most informative reading on the page.
→ How the Benchmark card is calculated · Everything in v0.1.34