How this hypothetical mix would have done
What am I looking at?
- Annual return
- The smoothed yearly growth rate (often called CAGR). 8% a year roughly doubles money every 9 years.
- Worst drop
- The largest fall from a previous high to a later low, also called maximum drawdown. This is the number that makes people sell at the bottom.
- Volatility
- How bumpy the ride was, measured as the yearly standard deviation of returns. Higher means bigger swings in both directions.
- Sharpe
- Return earned above cash, divided by volatility. It asks: how much reward per unit of bumpiness? Above 0.5 is decent for a diversified mix.
- Worst 12 months
- The worst result for anyone who started on the wrong day and checked a year later.
- Longest underwater
- How many months the portfolio spent below its previous high. Time in the hole tests patience as much as the size of the hole.
- Hypothetical value today
- What the starting cash plus monthly adds would have grown to under these rules. Real accounts would differ because of taxes, timing and behavior.
The lesson: Average return gets the attention, but worst drop and time underwater decide whether a real person sticks with a plan. A plan you abandon in a crash earns nothing like its backtest.
Try this: drag Risk tolerance from Moderate to Very high. Note how much Annual return rises compared with how much deeper the Worst drop goes.
Growth and drawdowns
How to read these two charts
Top chart: $10,000 growing over time. The blue line is the mix you built; the dashed gold line is the anchor alone. The scale is logarithmic, so a move from $10k to $20k looks the same size as $20k to $40k. That keeps early crashes visible instead of squashed flat.
Bottom chart: the drawdown, meaning how far below its last high the portfolio sat each month. Zero means at a new high. The deeper and wider each valley, the harder it was to live through.
The lesson: Two portfolios can end at a similar place and feel completely different on the way. The bottom chart is the experience; the top chart is the outcome.
Try this: set Cut rule to “Anchor to cash below its 10-month average” and watch the 2008 valley in the bottom chart get shallower. Then check what it did to the 2011 and 2022 valleys.
Growth of your starting cash only (no monthly adds) so rules compare fairly. Log scale, so equal vertical distance means equal % change.
Same weights, different rules: what changes?
What do these rules mean?
- Rebalance
- Selling some of what grew and buying what lagged to return to the target weights. It keeps risk from drifting upward as stocks rise.
- Calendar
- Rebalance on a set date, here once a year in December.
- Band
- Rebalance only when any holding drifts a set number of points from its target, for example 5 points. Acts on need, not the calendar.
- Trim into strength
- Sell only the part of a winner that is stretched well above its 10-month average and over its target, and move it to laggards. Lets winners run until they are extended.
- Cut rule (trend)
- Move a holding to T-bills when its month-end price closes below its 10-month average, and buy back when it closes above. Insurance against long bear markets, with a cost in choppy years.
- Calmar
- Annual return divided by worst drop. Higher means more return for each unit of pain.
The lesson: Rules mostly reshape the ride, not the destination. Look for the rule that keeps the worst drop inside your limit with the fewest trades, since every trade can mean costs and taxes.
Try this: compare “Never rebalance” with “Band 10 pts”. The difference in worst drop is the price of letting a portfolio drift.
Calmar = annual return ÷ worst drawdown (higher means more return per unit of pain). Trades per year counts full rebalances; trims are partial sales of an extended winner. Your selected rule is highlighted.
How a disciplined investor writes cut / add / trim rules
These are the illustrative rules tested in this lab, written out the way you would write your own. They show the format and the trade-offs; they are not instructions for your money.
Why write rules at all?
Most investing mistakes happen in the moment: selling after a crash, buying after a run-up, doing nothing because the decision feels too big. A written rule turns an emotional decision into a check: did the condition happen, yes or no?
A good rule names the trigger (what has to happen), the action (what you do), and the reason (what job it serves). Each card below follows that pattern, with the numbers from the mix you built.
The lesson: The best time to decide what you will do in a 30% drop is before one starts.
Why a 10-month average?
Every trend rule in this lab compares the month-end price with the average of the last 10 month-end prices. The number is a convention, not a law, so here is where it comes from and how much it matters.
It was popularized by Mebane Faber's 2007 paper A Quantitative Approach to Tactical Asset Allocation, which tested a 10-month simple moving average across asset classes back to the early 1900s. Ten months is roughly the monthly version of the 200-day average that institutional traders have watched for decades.
Checking once a month on the closing price ignores most of the day-to-day noise that makes faster rules flip back and forth. Fewer flips means fewer trades, lower costs and less tax in a taxable account.
A shorter average gets out sooner in a crash but whipsaws more in choppy markets. A longer one trades less but exits later and re-enters later. Ten months sits in the middle of that range.
If only one exact setting worked, the rule would probably be fitted to the past. The table below runs the same rule on your anchor with 4 to 14 months. Look for a plateau of similar results, not a single peak.
Crash replay: live through every drawdown
Press play to watch two portfolios fall below their highs and recover, month by month. The replay slows inside each crash so you can read what it teaches. Your full mix can only be tested from July 2007, when the bond and gold funds it uses all existed; switch to Long history to replay the 2000 to 2002 dot-com crash with the anchor alone, with and without the 10-month cut rule.
Every crash, side by side
Depth is the worst month-end fall inside each window, measured from the high just before that crash. Recovery counts months from that low until the pre-crash high was regained. The chart above measures from the all-time high, so it also shows when a portfolio was still underwater from an earlier crash. Best and worst asset are total moves inside the window.
The math that makes drawdowns so expensive
Trimming into strength, step by step
Watch one holding's price against its 10-month average (grey) and the trim line above it (dashed gold). Below, see its share of the mix with no rule (red) versus the trim rule (blue). Gold dots mark each trim.
The rule sells only the slice above target, and only when price is far above trend. The core holding stays, so you keep participating if the run continues.
Trims happen when a holding is extended and sentiment is warm, the opposite of selling in a panic. The cash lands in whatever lagged.
Gains moved into bonds, gold or cash during good times are what you rebalance from when the next drawdown arrives.
In a taxable account each trim can realize a gain. Many investors trim inside tax-advantaged accounts first, or redirect new contributions instead of selling.
Satellite scorecard
Core, satellites and the words in this table
- Core and satellite
- A large, cheap, steady core (the anchor and diversifiers) plus small satellite sleeves that try to add return, income or protection. Satellites are carved out of the total, so the core shrinks to make room.
- Alpha engine
- A rules-based sleeve that tries to beat the market, here by rotating into the strongest sectors, countries or large stocks each month.
- Momentum
- The tendency for assets that rose over the last 3 to 12 months to keep rising for a while. The engines rank by the average of 3, 6 and 12-month returns.
- Dividend fund
- A fund of companies that pay or grow dividends. Often a smoother ride, not always a higher return.
- Hedge
- A sleeve meant to rise, or fall less, when stocks crash. Hedges usually cost money in calm years; that cost is the insurance premium.
- Correlation
- How closely two things move together, from 1 (in lockstep) to -1 (opposite). Low or negative correlation to the S&P is what makes a hedge useful.
- Survivorship bias
- Testing only the companies that are winners today. They look brilliant in hindsight because the losers were dropped from the list. The Mag 10 rows carry this bias.
The lesson: A satellite earns its place by doing a job the core cannot, not by having the highest number in one column. Check the crash columns: a hedge that never rose in a crash is not a hedge.
Try this: set Hedge to TAIL at 5% and compare its crash columns with its annual return. That gap is what tail protection costs.
Each sleeve is measured from its own start date against the S&P 500 over the same months, with dividends reinvested. Engines decide at each month-end close and hold for the next month, so nothing peeks ahead.
How the engines rank, using the 31 Oct 2026 close
This shows how a monthly ranking rule turns prices into a decision, using one dated example. It is a teaching illustration, not a current signal or recommendation, and it goes stale the moment the next month closes.
How a three-lens stock screen works: fundamentals, sentiment, technicals
What each lens measures
- Fundamentals
- Is the business healthy? Uses the Piotroski F-score (nine yes/no checks on profitability, debt and efficiency, 0 to 9) and the PEG ratio (price-to-earnings divided by expected growth; under about 1.5 suggests you are not overpaying for growth).
- Sentiment
- What do professional analysts think? Uses the consensus rating. Useful context, but crowds can be wrong together.
- Technicals
- What is the price doing? Uses the monthly technical rating and whether the price sits above its 200-day average, a common trend line.
- The weights
- The sliders change how much each lens counts. Turning one up shows how much a screen's “answer” depends on the screener's own choices.
The lesson: A screen narrows a universe into a watchlist; it does not tell you what will go up. Today's ratings cannot be tested on the past, so this table shows a method, not evidence.
Try this: set Technical weight to 0 and Fundamentals to 3. Count how many names change. That is how sensitive a screen is to its design.
Each lens scores 0 to 3. Fundamentals: Piotroski F-score 7+ = 2 points, 5 to 6 = 1, plus 1 if PEG is between 0 and 1.5. Sentiment: analyst consensus Strong Buy 3, Buy 2, Neutral 1. Technicals: monthly technical rating Strong Buy 2, Buy 1, plus 1 if price is above its 200-day average. Universe: the five largest S&P 500 names in each TradingView sector. Fundamentals and analyst ratings are today's values, so this cannot be backtested honestly; use it as a watchlist and let the position rules below decide entries and exits.
Example position rules a screen needs before it can be used
Score 7 or more out of 9 at equal weights, monthly close above its 10-month average, and room under the cap: 2 to 4% of the portfolio per name, at most 3 names per sector, all single stocks together inside the Alpha 2 budget.
When it closes a month 25% or more above its 10-month average, or grows past 1.5 times its starting size, sell back to the starting size. Same logic as the portfolio trim rule, tuned wider because single stocks swing about twice as much.
Exit on a monthly close below the 10-month average, or if the score drops to 4 or less at the quarterly refresh. No averaging down on a name that has broken trend.
Refresh the screen once a quarter, after earnings season. Replace a holding only if the new name scores at least 2 points higher, which keeps turnover and taxes down.
Anchors: Nasdaq-100 vs S&P 500 vs total market, trading vs buy and hold
Why the start date changes everything
The same index can look brilliant or painful depending on when the test starts. Starting in March 1999 puts the Nasdaq-100 right before the dot-com collapse; starting in 2010 skips both the dot-com crash and 2008.
“Trend” here is one simple rule: hold the index when its month-end price is above its 10-month average, otherwise hold T-bills. It makes one decision a month and never uses information it would not have had at the time.
The lesson: Whenever you see a backtest, ask: why did it start there? A different start date is often a different conclusion.
Try this: compare the Nasdaq-100 buy-and-hold rows that start in Mar 1999 and Jan 2010. Same fund, very different story.
"Trend" holds the index when its month-end close is above its 10-month average and sits in T-bills otherwise (one decision a month, no look-ahead). Start dates matter: from 1999 the Nasdaq-100 fell 81% and took 14 years to recover.
Crash-by-crash returns
Same index, different wrapper: what a ticker actually changes
What is a wrapper?
- Index
- The list of stocks being tracked, such as the S&P 500.
- Wrapper
- The fund that holds that list for you. Several funds can track the same index, so they carry the same market risk.
- Expense ratio
- The yearly fee as a percentage of what you hold. 0.20% means $20 a year per $10,000.
- Liquidity
- How easily and cheaply a fund trades. Matters for frequent traders and options users, far less for long-term holders.
The lesson: Choosing between funds on the same index is a cost and trading-features decision, not a risk decision. Small fee gaps compound over decades.
Testing fund vs. holding fund
The tickers in this lab stand in for asset classes. Each one was chosen because it is the oldest liquid fund in its class, so the backtest can reach back through 2008 and earlier. Many cheaper funds tracking the same assets launched later, which means they cannot be tested that far back, even though they do the same job going forward.
Why history decides the testing fund
A backtest is only as long as its shortest-lived holding. If the gold sleeve used GLDM, launched in 2018, the whole portfolio could only be tested from 2018: no 2008 crash, no 2011, no 2015 to 2016, no Q4 2018. Using GLD from 2004 keeps every crash in the lesson. The asset class (gold) is what the lesson is about; the wrapper is a cost decision.
The lesson: pick the asset class first, then the cheapest, most suitable fund that holds it. The ticker that tested best is rarely the one built for holding.
| Asset class | Lab tests with | Since | Its fee / yr | Lower-cost funds, same job | What differs | Who might still use the original |
|---|---|---|---|---|---|---|
| US large companies (S&P 500) | SPY | 1993 | 0.0945% | VOO 0.03% · SPYM 0.02% | Same index, same holdings. | Active traders and options sellers: SPY has the deepest options market. |
| Nasdaq-100 | QQQ | 1999 | 0.20% | QQQM 0.15% · IQQ 0.10% | Same index, same holdings. IQQ is new, so it has little trading history. | Active traders and options sellers. |
| Total US market | VTI | 2001 | 0.03% | Already low cost | Nothing to swap. | Everyone. |
| Medium-term Treasuries | IEF | 2002 | 0.15% | VGIT 0.03% · SCHR 0.03% | These hold 3 to 10-year bonds, slightly shorter than IEF's 7 to 10, so they move a little less when rates change. | Investors who want IEF's exact maturity range. |
| Long-term Treasuries | TLT | 2002 | 0.15% | VGLT 0.03% · SCHQ 0.03% | These hold bonds of 10+ years, shorter on average than TLT's 20+, so they swing a bit less. | Traders who want maximum rate sensitivity and deep liquidity. |
| Gold | GLD | 2004 | 0.40% | GLDM 0.10% (2018) · IAU 0.25% (2005) | All hold physical gold; same tax treatment as collectibles. | Options traders: GLD's options market is far deeper. |
| Broad commodities | DBC | 2006 | 0.85% | PDBC 0.59% (2014) | Nearly identical futures basket. DBC issues a K-1 tax form; PDBC does not, but its structure is taxed differently, so check with a tax professional. | Few long-term holders. |
| US real estate (REITs) | VNQ | 2004 | 0.13% | SCHH 0.07% | SCHH holds REITs only and has paid a lower yield; VNQ is broader. | Income-focused holders who want VNQ's breadth and higher yield. |
| International developed | EFA | 2001 | higher | VEA 0.03% · IEFA 0.07% | VEA includes Canada and more small companies; IEFA excludes Canada. | Few long-term holders. |
| Cash (T-bills) | BIL | 2007 | T-bill rate | SGOV and other T-bill funds | Very short Treasury bills either way. | Either works. |
Fees are annual expense ratios as published by fund issuers and fund-research sites, checked November 2026; years are fund launch years. Fees change, so confirm them before acting. Funds named here are common examples of each asset class, shown for education, not recommendations. Satellites follow the same logic: the sector engine uses the original nine SPDR sector funds and the country engine ten long-running iShares country funds, because fixed, long-lived lists avoid hindsight in choosing them.
What actually diversifies
How to read the heat map
Each cell is the correlation between two assets' monthly returns from 2007 to 2026. 1.00 means they moved in lockstep; 0 means no relationship; negative means they tended to move in opposite directions.
Diversification is not owning many tickers. It is owning things that do different jobs, so that something holds up when the anchor falls.
The lesson: Three stock index funds with 0.9+ correlation are one bet wearing three names.
Try this: find the row for SPY and look for the lowest numbers. Those are the assets that actually changed the ride in 2008.
What the mix looks like in dollars
This simply splits the cash amount you entered by the weights above so the percentages feel concrete. The funds shown are common examples of each asset type (the testing fund, then a lower-cost alternative from the table above), not picks for you.
Lump sum or a little at a time?
- Lump sum
- Investing the full amount at once. Historically it came out ahead more often, because markets rise more often than they fall.
- Dollar-cost averaging
- Spreading the amount over several months. It usually costs a little return, but it reduces regret if a crash arrives right after you start.
The lesson: The “right” answer depends partly on behavior: the method you will actually stick with beats the one that wins on paper.
What wealth managers and institutions do that you can do too
Pension funds, endowments and private wealth desks win less by picking stocks than by process: written rules, risk limits, liquidity planning and cost control. Most of that process is available to an individual for free.
A written policy statement
Every pension and endowment runs on an Investment Policy Statement: objectives, risk limits, allowed assets, target weights, rebalancing ranges and who decides what.
Your version: one page answering the questions further down this lab. When markets get loud, the page decides, not the headlines.
Allocation before selection
Classic research on pension funds (Brinson, Hood and Beebower, 1986) found that asset allocation policy explained most of the ups and downs in their returns over time. Institutions set the mix first, then pick managers.
Your version: decide anchor, diversifiers and limits before choosing any fund or stock. That is the order of the controls on the left.
Rebalancing ranges, funded by cash flows
Large funds set a range around each target and act only when a weight leaves it, using incoming contributions first so they sell as little as possible.
Your version: a band rule plus new money sent to the most underweight holding.
A liquidity reserve
Pensions match the next few years of payouts with safe, short-term assets so they are never forced to sell stocks at the bottom to pay benefits.
Your version: keep the cash you may need soon outside the risk portfolio, in T-bills or a high-yield savings account. A forced seller cannot follow any rule.
Risk budgets and position limits
Risk teams size positions by how much risk they add, not by dollars, and cap any single manager, sector or name.
Your version: inverse-volatility weights (what this lab uses for diversifiers), a cap on every satellite, and single stocks kept small.
Stress tests before the stress
Investment committees run portfolios through past crises and hypothetical shocks, and agree in advance what they will do in each.
Your version: the crash replay and stress table. Write down your action for a 20% and a 35% fall before either happens.
Tax-aware implementation
Private wealth desks place assets in the account where they are taxed least, harvest losses to offset gains, and rebalance with new cash where they can.
Your version: the same three habits, checked with a tax professional for your own accounts.
Fierce cost control
Big funds negotiate fees down to a few basis points because costs compound against them every year.
Your version: the cheapest share class for long-term holdings. See the wrapper table for what a few hundredths of a percent add up to.
Return streams that zig
Endowments add assets that behave differently from stocks, such as trend-following, real assets and commodities, to smooth the ride.
Your version: gold, commodities and managed-futures ETFs now offer some of this in a brokerage account. Private equity and venture remain largely out of reach and illiquid.
Borrowing instead of selling
Wealthy clients often borrow against a portfolio through a securities-based line of credit rather than sell and realize gains.
Your version, with care: it only works at low loan-to-value. In a deep drawdown the lender can demand cash or sell your holdings at the worst moment. Study it before using it.
Governance and review rhythm
Committees meet on a schedule, follow a checklist and record why each decision was made.
Your version: a quarterly review date and a short decision journal. Over time it shows whether your rules or your moods drive results.
What you can do that they cannot
No quarterly reporting pressure, no career risk, no committee to persuade. You can hold cash without being fired for it, wait years for a plan to work, and use tax-advantaged accounts built for individuals.
The lesson: borrow the institution's discipline, keep the individual's patience.
What this lab cannot tell you
Backtests show what rules did in one history. Interest rates, inflation, valuations and market leadership can all behave differently from here.
Selling in a taxable account can trigger capital gains. Rules that trade often look better here than they would after tax. Retirement accounts behave differently.
The simulation never panics, never skips a month, and never second-guesses a rule. Real investors often earn less than their funds because of buying and selling at the wrong moments.
Some choices (which assets to include, the inverse-volatility weights, the Mag 10 list) were made knowing what happened. The more a rule is tuned to the past, the less it may hold up.
Someone adding money for decades and someone withdrawing in retirement can get very different outcomes from the same average return, depending on when the bad years land.
Monthly closes hide deeper intramonth drops. The real worst moment was usually worse than the number shown.
Questions to answer for yourself first
A portfolio is personal. These are the questions a good plan answers before any ticker is chosen. Write your answers down; they become your own rule book.
- What is this money for, and when will I need it?
- What is the largest drop I have actually lived through without selling, not the one I think I could handle?
- How much cash do I keep outside the portfolio for emergencies, so I am never forced to sell in a crash?
- What exactly will I do if the portfolio falls 20%? 35%? Who or what decides?
- How often will I review it, and what has to happen for me to change anything?
- Which account types am I using, and how do taxes change which rules make sense?
- Do I understand every holding well enough to explain its job in one sentence?
For decisions about your own situation, consider a qualified, licensed financial professional.
Check your understanding
Glossary
- Anchor
- The core stock index a portfolio is built around, such as the S&P 500.
- Diversifier
- An asset added because it behaves differently from the anchor, such as Treasuries or gold.
- Drawdown
- The fall from a previous high to a later low, shown as a percentage.
- CAGR
- Compound annual growth rate: the steady yearly rate that gets you from start to end value.
- Volatility
- How much returns swing around their average, scaled to a year.
- Sharpe ratio
- Return above cash per unit of volatility.
- Calmar ratio
- Annual return divided by the worst drawdown.
- Rebalancing
- Trading back to target weights after markets move them.
- 10-month average
- The average of the last ten month-end prices; a slow trend line. Price above it is read as an uptrend.
- T-bills
- Short-term US government debt; the lab's version of cash. Funds like BIL and SGOV hold them.
- Inverse-volatility weighting
- Giving calmer assets more weight and jumpier ones less, so each contributes similar risk.
- Momentum
- The tendency of recent winners to keep winning for a while.
- Managed futures
- Funds that follow trends across stocks, bonds, currencies and commodities, long or short. They often did well in 2022.
- Expense ratio
- A fund's yearly fee as a percentage of assets.
- Total return
- Price change plus dividends reinvested.
- Look-ahead bias
- Accidentally using information a real investor would not have had yet. This lab decides at each month-end close and acts the next month to avoid it.
- Survivorship bias
- Testing only today's survivors and winners, which flatters results.
- Backtest
- A simulation of rules on past data. Useful for learning trade-offs; never a guarantee.
Method and limits
Satellites (sector, country and Mag 10 engines, dividend funds, VXUS, EEM and hedges) are total returns with dividends reinvested, computed inside TradingView with Pine Script on 15 Nov 2026; the modeled S&P series used for the core matched the TradingView total-return S&P to within 0.1% a year. Engines: momentum = average of 3, 6 and 12-month return; a holding must close above its 10-month average; empty slots sit in BIL. Stock screen values are a 15 Nov 2026 snapshot. Core data: monthly closes for QQQ (from 1999), SPY (1993), VTI (2001) and TLT, IEF, GLD, DBC, VNQ, EFA (from Jul 2007) pulled from TradingView on 15 Nov 2026; the Nov 2026 bar is partial. Prices are split-adjusted but exclude dividends, so a long-run average distribution yield is added back each month (S&P/VTI 1.9%, Nasdaq-100 0.8%, long Treasuries 3.0%, 7-10y 2.4%, REITs 4.0%, intl 2.9%, gold 0). Cash and commodity collateral earn the yearly average 3-month T-bill rate. Wrapper results (QQQM, IQQ, VOO, SPYM) are modeled from the parent index with each fund's expense ratio; real tracking differs by a few hundredths of a percent. Monthly closes understate intramonth drawdowns (the S&P fell about 55% close-to-close daily in 2008 vs 51% here). Costs: 5 bps per dollar traded; taxes are not modeled, and trading in a taxable account adds drag. Diversifier weights use inverse volatility over 2007 to 2026, which looks back. Past results do not predict future returns. Educational analysis by Dr. Radiance Trades, not individual financial advice.
Want to understand the decisions behind the numbers?
Dr. Radiance Trades teaches portfolio architecture and decision-making frameworks so you can make your own informed choices. Book a 15-minute call with Dr. Sami to see whether DRT education fits you.