Curriculum·S210 Automation Literacy: What Bots Actually Do·about 31 min
When automation is genuinely the right tool
By the end of this lesson you can
- →Identify the three tasks automation reliably improves, and what they have in common
- →State the tasks it reliably worsens, and why the improvement is illusory
- →Apply the deployment checklist before any automation touches real funds
- →Choose the simplest automation that removes a named human failure
Sophomore · enrolled learners
This lesson opens with The one automation with no regime bet.
- What happened
- This is a structural finding rather than a named incident, and it is labeled as such. Every automated strategy examined in this course embeds a bet on a market state and fails in the opposite one: grid and mean reversion in a trend, trend following in a range, market making in volatility. Dollar-cost averaging is the exception. It buys a fixed amount on a fixed schedule, so it makes no claim about the state, contains no parameter that can be optimized toward a sample, and has no configuration space to search in the sense the backtest overfitting result describes. Its rule adds no failure mode of its own: if the asset falls, the schedule loses what the asset loses, and against a lump sum its usual cost is a lower return, since Vanguard found immediate investment ahead in about two-thirds of rolling 12-month periods in three markets. In our reading of the category, it is also the least marketed automation.
- The decision point
- The strategies with the largest marketing budgets are the ones with parameters, because parameters produce configurations, configurations produce impressive backtests, and impressive backtests produce customers. The one automation that removes a documented human failure without adding a regime bet has no backtest to advertise, which in our reading is why it is sold, where it is sold at all, as a feature rather than a product.
What you will be able to answer
- →What do the tasks automation improves have in common?
- →Why is dollar-cost averaging the exception?
- →What is the deployment checklist?
- →What is the test for adding automation?
Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.
It is free. We do not sell the list and there is nothing to buy at the end of it.
Sources and review
- https://www.dalbar.com/PressReleases/doc/2025QAIBPressRelease.pdf
- https://www.dalbar.com/PressReleases/doc/QAIB2024_PR.pdf
- https://www.kitces.com/blog/does-the-dalbar-study-grossly-overstate-the-behavior-gap-guest-post/
- https://web.archive.org/web/20200313144952/https://personal.vanguard.com/pdf/ISGDCA.pdf
- https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2308659
- https://onlinelibrary.wiley.com/doi/abs/10.1111/0022-1082.00226
- https://3commas.io/blog/api-security-incident-faq
Confidence medium·Volatility low·Reviewed 2026-08-06·Owner unassigned
Contested
The claim that dollar-cost averaging has no regime bet is structural rather than empirical. Whether it outperforms a lump sum is a separate question; Vanguard's 2016 study found immediate investment ahead about two-thirds of the time across the United States, the United Kingdom and Australia. That study is about phasing in a lump sum, not about buying from income as it arrives, which is the scheduled buying this lesson recommends; Vanguard draws that distinction itself.
DALBAR's investor-gap figures are disputed. Harry Sit, in a 2012 guest post on Kitces.com, argues that comparing dollar-weighted investor returns with an index confounds behavior with the sequence of returns and the timing of contributions, and so overstates the gap; DALBAR defends its method. The 848 basis points is a single year (2024), the second-largest gap of the past decade; the 2023 gap was 5.5 points. Neither is an annual rate, so the worked example uses an assumed 2 points a year and says so.
That dollar-cost averaging is the least marketed automation is our reading of the category, not a measured finding.
P10 applies with force to this lesson. Nothing here implies that any automation produces a return. The argument is that some automations remove documented human failures and most add regime bets.