A friend sent me a spreadsheet a while back with the subject line "smarter DCA," and by the third message he was asking whether it was normal that the sheet wanted him to deposit four months of contributions in one go. He had found value averaging. More precisely, he had found the half of value averaging that shows up in every backtest thread, without the half that shows up in your bank account during a real bear market. Both halves are worth understanding, because the strategy is genuinely good, the failure mode is genuinely ugly, and there is a middle version that keeps most of the good part.
What value averaging actually does
Dollar cost averaging you already know. Fixed amount, fixed schedule, no decisions. Put in $500 every month and the arithmetic quietly buys more units when price is low and fewer when it is high, because the dollars are constant and the price is not.
Value averaging, which Michael Edleson wrote the book on decades ago, flips the fixed part. Instead of fixing the contribution, you fix a target for what the portfolio should be worth at each point in time, and the contribution becomes whatever closes the gap. Say your path is $500 of value per month. After month four the portfolio should be worth $2,000. If a dip has your holdings at $1,400, you contribute $600 this month instead of $500. If a rally has you at $2,300, you contribute nothing, and strict value averaging would actually have you sell $300.
The effect is that your money leans against price twice. DCA only gets the quantity effect, more units per fixed dollar when price drops. Value averaging also scales the dollars themselves, so cheap months get more money and expensive months get less or none. You end up buying low and selling high on a schedule, with zero forecasting involved. One implementation detail matters here: grow the target path at a rough long-run expected return for the asset instead of using a flat line, otherwise ordinary appreciation keeps triggering sells of something you want to hold.
Where it beats DCA
In choppy, mean-reverting, sideways-with-violence conditions, value averaging historically posts a higher internal rate of return than plain DCA, and it does so fairly consistently in the studies I have read. The intuition is that every oscillation is a small round trip, and value averaging systematically puts more money in near the bottom of each wiggle and less near the top. Volatile assets that chop hard around a slow trend, which describes most crypto majors most of the time, are close to ideal terrain for it on paper.
Two caveats before anyone gets excited. First, the margin is typically modest, and IRR comparisons between the two are slippery, because the cash flows have different shapes, so the higher rate is being earned on different amounts of money at different times. Second, in a strong sustained uptrend value averaging underperforms, since it keeps you under-invested and trims your winner all the way up, with a taxable event on every forced sell if this lives outside a sheltered account. You are paying for a mean-reversion edge with trend-following pain.
If that were the whole tradeoff I would call it roughly a wash, slight edge to value averaging on volatile assets, and this would be a much shorter post.
Where it blows up
The contribution formula has no ceiling. Your required deposit is the gap between the target path and reality, and in a deep bear market that gap compounds against you every period. Walk through a rough example. Two years in, your path says the portfolio should be worth $12,000 and your holdings sit at $12,500, so there is nothing to do. Then the asset draws down 60% over the following months, which for crypto is a routine bear rather than a black swan. The path has stepped up toward $14,000 while your holdings have fallen under $6,000, and the sheet now asks for more than a year of normal contributions in a single month. Feed it, and the units you just bought can keep falling too, so the gap partially regrows while the path steps up again.
In a long grinding drawdown the demands can run at several multiples of your base contribution for many months in a row. There is no natural limit, because the formula quietly assumes an infinitely deep cash reserve. Edleson's own treatment handles this with a large side fund parked in cash or bonds, which works, and which also drags on your blended return the whole time markets are rising.
What happens to real people is one of two things. Either you run out of investable cash near the bottom, which stops the strategy at precisely the moment its entire edge is concentrated. Or you stare at the demanded number, decide the sheet has lost its mind, and quit. Both outcomes leave you worse off than boring DCA, because DCA never asks you a question you cannot answer. A mechanical strategy only counts as mechanical if you can keep executing it under stress, and unbounded cash calls are how this one fails that test.
The capped hybrid
The fix is to keep the value path but bound the cash flows. You trade away some theoretical edge for a system that survives contact with an actual bear. Here is the version I would hand that friend.
- Set a base amount B, whatever you would have comfortably DCA'd anyway. That number should already survive a bad income year.
- Build the target path from B and grow it at a rough long-run expected return, so normal appreciation does not trigger sells.
- Each period, compute the raw value-averaging contribution, target minus actual. Then clamp it: never below zero, never above roughly two to three times B.
- Skip the selling side entirely in taxable accounts. When you are above path, contribute nothing and route that month's B into a cash buffer instead.
- Hold the buffer somewhere stable and yield-bearing, sized around several months of the capped maximum, and make it the only source for above-B contributions.
- Re-anchor quarterly. If the gap has grown beyond what a few capped months can close, reset the path to start from the current actual value instead of trying to catch up. This single rule is what removes the unbounded demand.
What the cap costs you is the very deepest buys at the exact bottom, which is where strict value averaging earns a chunk of its historical outperformance. What it keeps is the whole buy-more-when-cheap tilt inside a range you can actually fund, plus the restraint of contributing less when the market runs hot. In practice the capped version behaves like DCA with a volatility bonus in normal conditions, and simply behaves like DCA in the conditions where the strict version behaves like a margin call from your own spreadsheet.
Before running this on anything, backtest the capped rules against plain DCA through the worst drawdown in that asset's history, not a typical one, and pull out two numbers: the largest single monthly cash demand, and the longest streak of consecutive above-B months. Side-by-side questions like that are part of why we built the backtester at Blockcircle, though a spreadsheet and some candle history will get you most of the way.
If that worst month is an amount you can wire without wincing, you have found your cap. If it makes you flinch, lower the multiple and rerun, because the only version of this that pays is the one you are still executing at the bottom.