| The Great ETF Explosion |
Currently there are over 5,000 US-listed ETFs trading across the major US exchanges – more than twice as many as there are companies listed on the NYSE.
What started out as a way of capturing the large market movement has morphed into thousands of ‘baskets of stocks’ capturing different segments and combinations of stocks.
Why SPY?
With more than 5,000 ETFs to choose from, why does the ITM Plan still use SPY? It’s so expensive! Which means, of course, that even DITM options with very little time value are expensive too. So, I looked for cheaper alternatives.
The Shortlist
After searching, I found only four US-listed ETFs close enough to SPY to examine, together with Mini-SPX index options. There weren’t any other obvious contenders. I started making a table of the alternatives.
- VOO: Vanguard S&P 500 ETF
- IVV: iShares Core S&P 500 ETF
- RSP: Invesco S&P 500 Equal Weight ETF
- SPYM: SPDR Portfolio S&P 500 ETF
- Mini SPX: Cboe Mini-SPX Index options.
Now to test if we can use them.
First Test: Does it track the same thing?
Well, RSP (Invesco S&P 500 Equal Weight ETF ) fell at the first hurdle.
Yes, it has all the S&P 500 stocks but they are equal-weighted, not weighted by market cap like the index. This makes the effect of the very large stocks like Nvidia and Microsoft very much less – the same, in fact, as all other stocks (at each rebalance). Is this a good idea? Possibly. Interesting, but maybe for another blog.
BUT – it does not mirror the S&P 500, therefore our backtesting cannot simply be transferred to it. One down.
Second Test: Is it Cheaper?
The prices when I checked on 30 Aug 2026 were:
- SPY: $771.10
- VOO: $708.75
- IVV: $774.75
- SPYM: $90.57
- XSP: $771.18
The obvious stand out is SPYM (formerly SPLG) at less than 12% of the SPY price. And options will be cheaper – let’s check them out.
The options are definitely cheaper. If we take the 50% option ($45 strike) then we only need $4,780 to get into the trade. If we choose the 60% option ($55 strike) we only need $3,800.
You’ll notice that I am estimating the capital required from the ask price rather than the midpoint of the bid and ask. Why? Because we have a liquidity problem here. Check out the spread: 8.6%. If we bought at the ask and immediately sold at the bid, we would lose 7.9%. The bid would have to rise by more than 8% merely to reach the price we paid.
The time value is also higher. Although this is a deep-in-the-money option, paying $38 for the $55 call gives us an effective purchase price of $93 when SPYM is trading at $90.57. The difference – $2.43, or 2.7% of the SPYM price – is time value.
So, yes, it is cheaper, but the disadvantage is the extra time value you have to pay for.
What about the other contenders?
XSP initially sounds promising because it represents one-tenth of the S&P 500 Index and has listed options. However, its quoted level is approximately the same as SPY’s share price, so it provides no meaningful capital advantage.
VOO and IVV are also priced relatively close to SPY. VOO offers a modest reduction in capital, but both have considerably less options liquidity.
Here are the figures from the comparable contracts:
The percentage spread is calculated as the spread divided by the bid price.
Why are the spreads wider?
The quoted option premium is only part of the cost. The spread determines how much we may lose simply by entering and exiting the trade.
The difference is largely liquidity. SPY attracts enormous trading activity and intense competition between market makers, producing deep markets and narrow spreads. VOO, IVV and particularly SPYM attract much less options trading, so their spreads tend to be wider and midpoint fills less certain.
XSP is different again. It is a proprietary Cboe index option supported by market-maker quotes. A displayed quote does not necessarily indicate active trading at that particular strike and expiry.
ETFs proliferate but liquidity concentrates
Thousands of ETFs have divided the market into ever smaller and more specialised baskets. But options liquidity does not spread itself evenly across all those products. It tends to cluster in a relatively small number of heavily traded contracts.
What does it mean for ITM?
SPY is not used because it is the cheapest fund, or because no other fund follows the S&P 500. It is used because it combines:
- the correct underlying index;
- a long and reliable price history;
- suitable long-dated options;
- enormous options liquidity;
- narrow spreads;
- efficient entry, exit and rolling.
For investors, there are thousands of ETFs. For the ITM Plan, there is still no obvious replacement for SPY.
Possibly with QQQ there are better contenders, but that’s for another blog.
Why did I choose options with strikes close to 60% of the underlying price and January 2027? Simply because I wanted the closest possible comparison. I used the same expiry – 15 January 2027 – and selected the nearest suitable strike available for each product.
To the markets . .
Not a brilliant week – although up for the week. Nvidia’s earnings were, of course, the big news, with revenue and profit exceeding Wall Street expectations.
Salesforce was also up nearly 23%. When I read that, I thought, “Well, how wrong can you be!” You may remember that I had lumped it in with SaaS (Software as a Service) stocks and said that the future did not look rosy for them. I was preparing to eat humble pie when I saw this headline:
So perhaps my theory wasn’t entirely shot out of the water. Salesforce had reported strong results and raised its forecasts – but it had also seen the writing on the wall and expanded its partnership with Anthropic. Smart.
So, my theory wasn’t shot out of the water – instead they saw the writing on the wall and made a partnership. Smart.
SPY Charts
SPY is still consolidating. Support at 760 is continuing to hold, but no new highs this week.
The long term chart shows the trend line is holding.
SPYG Charts
SPYG is also in consolidation and has returned to its upper bound. Hopefully this week it breaks through.
Long term it is still near the top of the trading channel.
QQQ Charts
QQQ is also still consolidating – as it has been for the last 4 months. Boring? Yes – but remember last October it was consolidating for almost 6 months – check out the second chart.
Longer term we see it is still crawling along the upper bound of its trading channel.
VIX Chart (Volatility)
VIX has popped up slightly but doesn’t seem significant.
ITMeter
The week ahead . . .
Employment will dominate the week, culminating in Friday’s August jobs report. Broadcom heads a busy technology earnings calendar.
Monday 31 August
Announcements: No major announcements.
Earnings: Science Applications International.
Tuesday 1 September
Announcements: Final Manufacturing PMI, ISM Manufacturing Index, JOLTS job openings, construction spending.
Earnings: Nio, Dell Technologies, Palo Alto Networks, MongoDB, GitLab, Credo Technology.
Wednesday 2 September
Announcements: ADP Employment Report, factory orders, Federal Reserve Beige Book.
Earnings: Broadcom, Snowflake, Hewlett Packard Enterprise, NetApp, C3.ai.
Thursday 3 September
Announcements: Weekly jobless claims, revised productivity and labour costs, trade balance, final Services PMI, ISM Services Index.
Earnings: Ciena, Lululemon, DocuSign, Zscaler, UiPath, Samsara, Asana.
Friday 4 September
Announcements: Nonfarm payrolls, unemployment rate, average hourly earnings.
Earnings: RH, Korn Ferry.
Friday’s employment report is the week’s main event. It will provide the clearest indication of whether the US labour market is continuing to weaken – and could materially alter expectations for the Federal Reserve’s next move. Broadcom’s results will also be closely watched for evidence that spending on AI infrastructure remains strong.
The Futures . . .
Not looking very nice. The Middle East situation seems to have escalated again, probably what is causing it.
Next week . . .
I have an announcement to make: my swansong wasn’t my swansong after all.
I have realised that books are no longer the best way for me to take ITM forward. The market is swamped with AI-produced books, leaving precious little room for actual human authors.
But people still need knowledge – they simply consume it differently. And that is where I am heading.
ITM may have a fantastic result, but I don’t expect anyone to accept it on trust. Everything I say is backed by data. I don’t conveniently discard anything that fails to fit.
So, I decided to make it easy for anyone to replicate the ITM results.
First, I updated the backtesting to 30 June 2026. Then AI rebuilt and checked the results using only the ITM rules and the raw data: date, opening price and closing price. The figures held up.
Now comes the real test: independent people replicating the results – or finding errors.
I have created a Replication Pack containing the data, rules, formulae and spreadsheets – the lot. Next week, I will invite you to download it and check the results for yourselves.
Think of it as beta testing before we go public.
So, onwards to the next stage of ITM!
Hoping for a great week!
Heather
Trade the tide not the waves.
4 Responses
Hello Heather,
Always good to read your blog. You never disappoint with any of your content.
I’ve been playing around with Claude recently just to evaluate its capabilities and thought I’d get its take on Dennis’s proposal re futures contracts, so I pasted in his proposal, and then pasted in your reply and the following is what I got back. It also included charts and tables that I can’t include here. Claude is pretty impressive, more so than ChatGPT I’d say.
Great connection to pull on — and this is very buildable with what we already have. Dennis’s core claim (“no time value, cleaner implementation”) and Heather’s concern (margin/mark-to-market risk) are both testable directly from our existing signal and data.
Modeling approach: I don’t have historical ES/MES futures data, but a fairly-priced, continuously-rolled S&P futures position is economically close to holding the index itself while posting only margin (not the full notional) — the “cost of carry” pricing largely cancels out, so the return profile should track the underlying’s total return closely, without the option’s extrinsic-value decay. That means the cleanest way to test this is to compare the 10/200 signal applied to shares (a fair proxy for fairly-priced futures P&L) against what we already built for deep-ITM calls — the difference between those two is roughly the cost Dennis is pointing at.
Then, separately, I’ll directly test Heather’s margin worry: what’s the worst peak-to-trough decline the position would have had to absorb while still in a trade, before the exit signal fired?
Ran 2 commands, created a file
I need to correct something before presenting this — my equity curve above only marked P&L at trade boundaries, not daily. Since Heather’s whole concern is about daily mark-to-market, that’s exactly the wrong way to measure this. Let me rebuild it properly.
Ran a command
Good — corrected: the true daily-marked drawdown is −20.3% (not the −8.3% my trade-boundary-only calculation showed), occurring in June 2020, which lines up closely with the −18.9% intra-trade figure I found looking within individual trades. That’s the honest number, and it directly answers Heather’s question.
Now let’s actually engage with what Dennis and Heather are debating.
1. Dennis’s core claim — “no time value, cleaner implementation” — checks out mechanically. A fairly-priced, continuously-rolled futures position tracks the underlying almost 1:1 with no optionality decay. Modeling it this way (using the 10/200 signal on shares as the proxy, since fairly-priced futures P&L should track total return closely):
This is the real finding: at matched 1x notional exposure, futures give you lower return and lower drawdown than the deep-ITM calls — because the DITM calls carry embedded leverage (you’re controlling a full share’s worth of exposure for only 50-60% of its price), which is where their extra return and extra risk both come from. Dennis’s letter undersells this a bit — futures aren’t just “cheaper,” they let you dial in exactly how much leverage you want (via MES’s $5 multiplier) without paying for it in decaying extrinsic value. If you sized futures contracts to match the DITM calls’ ~1.7-2x effective notional exposure, you’d likely get similar or better returns than the DITM version, without ever bleeding time value — that’s the genuine structural edge futures have, and it’s a more precise version of what Dennis is gesturing at.
2. Heather’s margin concern — this is answerable with real numbers, and it’s a fair worry. I checked the worst peak-to-trough decline the position would have had to sit through while still in a trade, before the trend-exit signal fired:
Worst case: −18.9% to −20.3%, right into the March 2020 COVID crash (the Feb 2019 → Mar 2020 trade). The 10/200 signal is inherently laggy — it didn’t get the account out before that decline happened, only after.
Second worst: −17.3%, during the 2011 debt-ceiling/European crisis scare.
The strategy did successfully sidestep almost all of the 2008 GFC (it was flat/out of the market for ~18 months spanning the worst of that crash) — but that’s a matter of timing, not guaranteed protection.
What this means for margin practically: if you were running this near 1x notional via futures, a margin/cash buffer sized to comfortably absorb a ~20-25% adverse move (with some room beyond the historical worst case, since 2020 isn’t guaranteed to be the worst future case) would likely have kept you clear of a margin call historically. If you leverage up further (say 2x notional, closer to matching the DITM calls’ embedded leverage), that same 18.9% underlying move becomes roughly a 38% hit to your margin account — which is where Heather’s instinct to be cautious is well-founded. Her stated preference for defined-risk-at-entry (DITM calls) over open-ended daily mark-to-market is a legitimate risk-management choice, not just “allergy” — it trades some of the futures approach’s cost efficiency for the certainty of a known maximum loss.
One honest caveat: this is proxied through SPY shares as a stand-in for fairly-priced futures, not actual ES/MES historical data — real futures would have small basis effects around quarterly rolls, and margin requirements themselves fluctuate (often increasing right when volatility spikes, which is exactly when you’d least want that). Heather’s instinct that this “would require a separate backtest incorporating contract sizing, cash reserves, and quarterly rolls” is the right call for anyone actually implementing this live — I’d treat what we’ve built here as a solid first-pass answer to “is this worth pursuing further,” not a final green light.
Want to push on the leverage-matching angle — size a futures-proxy position to roughly match the DITM calls’ effective exposure and see if it actually beats the options version once you strip out the time-decay cost?
Yes (me).
Applying those ratios (1.96x for the 50%-strike-equivalent exposure, 2.39x for the 60%-strike-equivalent) directly to the futures-proxy daily returns:
Ran a command
This settles it — at matched leverage, futures beat the DITM options on both fronts, confirming Dennis’s argument precisely.
At matched effective exposure, futures win on both metrics, every time. The gap (roughly +0.9 to +1.9 points of CAGR, and slightly shallower drawdowns) is exactly the cost Dennis was pointing at in his letter: even a deep in-the-money call still carries a sliver of extrinsic value that decays over the life of the option, plus our modeled volatility-risk-premium assumption (IV priced a bit above realized vol, which is realistic for how options actually trade). Futures don’t carry either of those costs — the P&L is purely linear against the underlying’s price move, so once you size the position to match the same effective leverage, you keep 100% of the return the leverage is generating instead of handing a small piece of it to time decay and implied-vol richness.
This is a clean, honest confirmation of Dennis’s core argument — futures aren’t just operationally “cleaner” (better liquidity, tighter spreads, flexible sizing via MES), they’re economically cleaner too: no optionality tax on the leverage you’re using. Heather’s counter — her preference for defined risk-at-entry over daily mark-to-market/margin exposure — is still the legitimate trade-off on the other side of this. The historical worst-case intra-trade decline we found earlier (−18.9% to −20.3% unlevered, so roughly −37% to −44% at this matched leverage) tells you concretely what a margin account would need to be sized to absorb. Futures buy you the better return/drawdown profile only if you’re comfortable managing that daily mark-to-market and margin risk directly — the DITM calls trade a small performance drag for the comfort of “I already know my maximum possible loss the moment I open the position.”
One thing worth flagging honestly: this is all still proxied through SPY’s own price history at a chosen leverage multiple, not real ES/MES contract data — actual futures would have small basis wobbles around quarterly rolls, and real margin requirements move around (often tightening exactly when volatility spikes, which could force you to de-lever at the worst possible moment even without a formal margin call). Heather’s instinct that this deserves its own dedicated backtest with real contract specs before going live is the right one.
Want to look at anything else here — e.g., stress-test what happens to the leveraged futures version if a future decline is meaningfully worse than the 2020 COVID case (a next-worst-case sensitivity test), or wrap up?
To be continued (me). Look forward to your feedback (me, not Claude).
Wow Eric – thats a lot to take in! I’d love to give feedback right away – but I am nose to the grindstone updateing the book and getting the Replication pack ready.
In a couple of weeks I should be able to life my head and looks at your work – it seems genuinely exciting – but I am being very strict with myself ATM – head down, no deviation.
But please keep me posted.
h
Hi Heather,
I enjoyed your latest piece on SPY versus SPYM. Have you ever considered using S&P 500 futures for the ITM concept instead?
We’ve been using futures for this concept for years. ES has a $50 multiplier and MES a $5 multiplier, so sizing can be very flexible. Compared with deep ITM options, they’re generally cheaper and simpler to trade, with excellent liquidity, very tight spreads and no time value.
The obvious drawback is margin, but with the strategy’s exit rules, you should normally be out well before margin becomes an issue in a major drawdown.
For systematic leveraged S&P 500 exposure, futures have always struck us as the cleaner implementation.
Curious whether you’ve looked at this before.
Best,
Dennis
Hi Dennis
Thank you—that is an interesting suggestion. No, I haven’t investigated using ES or MES for the ITM Plan.
I can see the attraction: excellent liquidity, tight spreads, flexible sizing with MES and no option time value. In principle, the same ITM signals could determine when to enter and exit the futures position.
My immediate reservation is that I am allergic to margin! One reason I have always preferred fully paid DITM calls is that the maximum loss is known at entry and there can be no margin call. With futures, the daily mark-to-market and the possibility of margin requirements increasing during a sharp decline would need careful modelling. I wouldn’t want to assume that the ITM exit signal would always arrive before margin became an issue.
It would therefore require a separate backtest incorporating contract sizing, cash reserves, daily gains and losses, margin requirements and quarterly rolls. But it is certainly a legitimate alternative—and one I had not previously considered.
Thank you for raising it.
H