{"id":100,"date":"2026-10-08T23:00:10","date_gmt":"2026-10-09T06:00:10","guid":{"rendered":"https:\/\/americanetp.com\/forum\/2026\/10\/08\/s-p-p-e\/"},"modified":"2026-10-08T23:00:10","modified_gmt":"2026-10-09T06:00:10","slug":"s-p-p-e","status":"publish","type":"post","link":"https:\/\/americanetp.com\/forum\/2026\/10\/08\/s-p-p-e\/","title":{"rendered":"S&#038;P P\/E: Trailing, Forward, Weighted"},"content":{"rendered":"<h2 id=\"what-is-the-s-p-p-e-ratio\">What is the S&amp;P P\/E ratio?<\/h2>\n<p>The S&amp;P 500 P\/E ratio is the index level divided by the index&#39;s earnings per index share, where that earnings figure is an aggregation of the reported or estimated earnings of every holding, scaled by the same divisor used to produce the index level. It is an output of two things: a dated holdings file and an earnings convention. Change either and the number changes, without any company trading at a different price.<\/p>\n<p>There is no single &quot;S&amp;P P\/E&quot;. The aggregation rule, the earnings basis, the forward horizon and the treatment of loss-making holdings are all choices made by whoever publishes the figure, and most published figures do not state which choices were made. Two desks can pull the same index level, the same holdings and the same earnings data and print numbers several points apart, each of them arithmetically correct.<\/p>\n<p>That is not a data quality problem. It is a specification problem, and it is solvable only by writing the specification down. The full arithmetic, including the divisor mechanics, is worked through in <a href=\"https:\/\/americanetp.com\/forum\/2026\/10\/06\/sp500-pe-ratio\/\">S&amp;P 500 PE Ratio: How It&#39;s Computed<\/a>; this article is about why the same arithmetic produces different answers and how to reconcile them.<\/p>\n<h2 id=\"why-two-desks-reading-the-same-index-get-different-p-e-numbers\">Why two desks reading the same index get different P\/E numbers<\/h2>\n<p>Four degrees of freedom fork the answer from identical inputs: the aggregation method, the earnings basis, the horizon, and how losses are handled. Each one moves the ratio in a predictable direction, which means a disagreement between two published figures can usually be attributed rather than argued about.<\/p>\n<p>The aggregation method matters most and is discussed least. Three rules are in common use, and only two of them agree with each other:<\/p>\n<ul>\n<li><strong>Aggregate ratio<\/strong>: sum of market capitalisation over sum of earnings. This is what most index publishers mean by the index P\/E.<\/li>\n<li><strong>Weighted harmonic mean of company P\/Es<\/strong>: mathematically identical to the aggregate ratio when the weights are market-cap weights.<\/li>\n<li><strong>Weighted arithmetic mean of company P\/Es<\/strong>: a different number entirely, and the outlier of the three, because any holding with earnings near zero produces an enormous individual P\/E that the weighted average then carries into the index figure.<\/li>\n<\/ul>\n<div class=\"ab-table\">\n<table>\n<thead>\n<tr>\n<th>Choice<\/th>\n<th>Direction it pushes the ratio<\/th>\n<th>Why<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Weighted arithmetic mean instead of aggregate<\/td>\n<td>Usually higher, occasionally wildly so<\/td>\n<td>Near-zero denominators create huge individual multiples<\/td>\n<\/tr>\n<tr>\n<td>GAAP instead of operating earnings<\/td>\n<td>Higher<\/td>\n<td>GAAP earnings include write-downs and one-off charges<\/td>\n<\/tr>\n<tr>\n<td>Forward instead of trailing<\/td>\n<td>Lower in most periods<\/td>\n<td>Estimates normally sit above the trailing twelve months<\/td>\n<\/tr>\n<tr>\n<td>Dropping loss-makers<\/td>\n<td>Lower<\/td>\n<td>Removes negative earnings from the denominator sum<\/td>\n<\/tr>\n<tr>\n<td>Including loss-makers at negative values<\/td>\n<td>Higher<\/td>\n<td>Negative earnings shrink the aggregate denominator<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>A worked example makes the spread concrete. Take a three-line index: holding A at a market cap of 600 with earnings of 30, holding B at 300 with earnings of 15, and holding C at 100 with earnings of negative 5.<\/p>\n<p>The aggregate ratio is 1000 divided by 40, or 25.0. Drop holding C and it becomes 900 divided by 45, or 20.0. The cap-weighted arithmetic mean of the three individual P\/Es, with C carried at negative 20, is 16.0. Three defensible methods, three answers, one set of inputs.<\/p>\n<p>Now change holding C&#39;s earnings from negative 5 to positive 0.5. Its individual P\/E becomes 200, the aggregate ratio falls to roughly 22.0, and the weighted arithmetic mean jumps to 38.0. A single holding at one tenth of index weight has moved one method by sixteen points and the other by three. That asymmetry is the reason the aggregate ratio, not the arithmetic mean, is the default in index work.<\/p>\n<h2 id=\"how-the-earnings-denominator-is-built-from-a-holdings-file\">How the earnings denominator is built from a holdings file<\/h2>\n<p>The denominator is built by pulling a dated holdings list with shares outstanding and float factors, mapping each line to an earnings series, applying the float-adjusted weight, and summing. The index level divided by that sum gives the ratio. Every step is mechanical except the mapping, which is where most reconstructions break.<\/p>\n<p>The procedure, in order:<\/p>\n<ol>\n<li>Fix an as-of date and pull the holdings file for that date, including shares outstanding and the float adjustment factor for each line.<\/li>\n<li>Map each holding to an earnings series using a stable instrument-level identifier, not a ticker.<\/li>\n<li>Choose the earnings basis and horizon, and apply it consistently to every line, including the ones you would rather exclude.<\/li>\n<li>Compute float-adjusted market capitalisation per holding and sum it.<\/li>\n<li>Sum the float-adjusted earnings across the same holdings, using the same share counts.<\/li>\n<li>Divide. The divisor cancels, provided both the numerator and the denominator use the same holdings snapshot and the same divisor.<\/li>\n<\/ol>\n<p>That last point deserves emphasis because it is the source of a common error. The divisor cancels only when both sides of the ratio are built from the same snapshot. Published index earnings-per-share series are updated on a reporting schedule while the index level updates continuously, so a ratio formed from a live index level and a stale per-share earnings figure carries whatever divisor change happened in between.<\/p>\n<p>Ticker-level mapping fails predictably on dual-listed names, share class pairs, post-merger survivors and recently reclassified lines, because a ticker is a venue-specific label that is reassigned. An instrument-level identifier such as a Bloomberg FIGI does not collide in those cases, which is why holdings files that carry FIGIs alongside the ticker remove an entire category of silent reconciliation failure. AmericanETP includes Bloomberg FIGIs in its holdings files alongside the ticker for exactly that reason.<\/p>\n<p>When the same exercise is run per sector rather than for the whole index, a classification layer sits between the holdings file and the sum, and the classification scheme becomes a fifth degree of freedom. <a href=\"https:\/\/americanetp.com\/forum\/2026\/08\/31\/etf-list-by-sector\/\">ETF List by Sector: 2026 Classification Guide<\/a> covers how those sector cuts are constructed and where two schemes assign the same company differently.<\/p>\n<h2 id=\"trailing-versus-forward-what-the-standard-chart-is-actually-plotting\">Trailing versus forward: what the standard chart is actually plotting<\/h2>\n<p>A forward P\/E chart plots a continuously updating numerator against a denominator that is a consensus of analyst estimates, and the standard presentation adds a long-run average and dispersion bands around it. Those bands are statements about a chosen distribution, not about valuation.<\/p>\n<p>The <a href=\"https:\/\/en.macromicro.me\/\">MacroMicro S&amp;P 500 Forward PE Ratio chart<\/a>, as read on 9 October 2026, plots the forward P\/E against a 25-year average with plus and minus one standard deviation bands and several moving averages. That construction makes two assumptions worth naming. The first is that the mean of the series is stationary across 25 years, which requires the index&#39;s sector composition and the estimate methodology to be comparable at both ends of the window.<\/p>\n<p>The second is that the estimate series itself is stable, and it is not. Analyst coverage, the number of contributors per name, the treatment of adjusted items and the revision cadence all change over a quarter-century, so the forward denominator in 2026 is not the same kind of object as the forward denominator in 2001.<\/p>\n<p>A forward P\/E also falls mechanically when analysts raise estimates faster than prices move, which happens routinely during an earnings upgrade cycle. A reader who interprets that fall as the market becoming cheaper has attributed to price something the denominator did. Band-crossing carries the same hazard: crossing plus one standard deviation says the current reading sits in the upper tail of the chosen 25-year sample, and nothing more.<\/p>\n<h2 id=\"sector-mix-when-the-index-p-e-moves-and-no-company-re-rates\">Sector mix: when the index P\/E moves and no company re-rates<\/h2>\n<p>The index P\/E can move materially while every single holding trades at an unchanged multiple. A rise in the weight of higher-multiple sectors lifts the aggregate ratio by arithmetic alone, as does a quarterly rebalance, an addition, a deletion, or a share-count update that changes the divisor.<\/p>\n<p>This is the composition effect, and it is the single most common reason a quarter-over-quarter P\/E change fails to match any intuition about valuation. A float update on a large holding changes its weight, changes the aggregate market capitalisation, and changes the aggregate earnings sum, all without a single trade.<\/p>\n<p>The diagnostic is a two-snapshot decomposition. Take dated holdings files from the start and end of the period, hold weights at their starting values while letting individual multiples move, and compute the resulting index P\/E: that is the within-name component. Then hold multiples fixed at their starting values while letting weights move to their ending values: that is the weight-shift component. The two components plus a small interaction term reconstruct the full change.<\/p>\n<p>Running that decomposition requires holdings snapshots you can actually retrieve for arbitrary past dates, which is a different requirement from having today&#39;s file. Sector-level P\/Es computed from an S&amp;P 500 sector ETF&#39;s holdings and from the underlying select sector index should reconcile closely, and when they do not, the usual causes are a different as-of date, the fund&#39;s cash and futures line, or a sampling difference between the fund and the index it tracks. Attributing the residual to specific index events requires a record of which additions and deletions landed between the two dates.<\/p>\n<h2 id=\"fund-pages-index-pages-and-the-p-e-you-see-published\">Fund pages, index pages and the P\/E you see published<\/h2>\n<p>A provider fund page and an index factsheet will often disagree on the same sector&#39;s P\/E, and the disagreement is structural rather than erroneous. A fund page reports a portfolio-level ratio computed from the fund&#39;s actual holdings on a stated date, including cash and any sampling difference. An index factsheet reports the index&#39;s ratio on the index&#39;s own schedule.<\/p>\n<p>The iShares S&amp;P 500 Consumer Discretionary Sector UCITS ETF page on <a href=\"https:\/\/www.blackrock.com\/\">BlackRock&#39;s site<\/a>, read on 9 October 2026, carries separate Portfolio Characteristics and Holdings sections. Those are precisely the two layers a reader has to reconcile: a published portfolio-level ratio on one hand, and the holdings that would have to reproduce it on the other. Nothing on a fund page obliges those two sections to share an as-of date, and often they do not.<\/p>\n<p>There is a second hazard in citing a vendor-hosted ratio rather than recomputing one. An ACR page on normalized earnings that still ranks for this query returned a 404 when fetched on 9 October 2026, which means any article that cited its figure now cites nothing. A number you can rebuild from a holdings file survives the publisher&#39;s site reorganisation; a number you linked to does not.<\/p>\n<p>That is the practical argument for keeping archived holdings rather than archived screenshots. AmericanETP holds archived holdings files back to July 2009, which is the raw material for re-deriving a historical index or sector P\/E on your own specification long after the page that originally published it has gone.<\/p>\n<h2 id=\"normalised-cyclically-adjusted-and-shiller-style-variants\">Normalised, cyclically adjusted and Shiller-style variants<\/h2>\n<p>The normalised family of ratios exists to suppress the earnings cycle, on the view that a trailing twelve-month denominator captured at a cyclical peak or trough is not a basis for comparison across decades. The three common approaches are a ten-year inflation-adjusted average of earnings, a peak-earnings normalisation, and a sales-based substitute such as price to sales.<\/p>\n<p>Each buys something and costs something. The ten-year average, the construction behind the cyclically adjusted ratio associated with Robert Shiller, removes almost all cycle noise and in exchange introduces a long lag, so it responds slowly to genuine structural changes in margin levels. Peak-earnings normalisation reacts faster but depends on an arbitrary choice of which peak. Price to sales sidesteps the negative-earnings problem entirely at the cost of ignoring margin structure, which is a large thing to ignore when comparing a 1996 index to a 2026 one.<\/p>\n<p>All of the multi-year variants share one flaw that is rarely stated. Averaging ten years of earnings for an index whose composition turned over substantially across the window compares two different portfolios and calls the difference a valuation signal.<\/p>\n<p>The honest version of the calculation splices archived holdings: for each year in the window, use the holdings as they stood in that year, or alternatively hold today&#39;s holdings fixed and backfill their own earnings histories. The two approaches answer different questions, and neither is the default output of a chart you downloaded. Doing either requires dated holdings going back the length of the window, which is the practical constraint that keeps most desks on the trailing measure.<\/p>\n<h2 id=\"e-mini-s-p-500-futures-and-the-index-p-e\">E-mini S&amp;P 500 futures and the index P\/E<\/h2>\n<p>E-mini S&amp;P 500 futures price the index level, not the earnings denominator. An overnight move in the e-mini changes the implied P\/E by exactly the price move and by nothing else, because the denominator was last updated on a reporting schedule and will not move again until the next earnings input lands.<\/p>\n<p>Futures fair value is a financing calculation: the spot index level carried forward at a short-term rate, less the dividends expected before expiry. Basis, the gap between the futures price and spot, is therefore a statement about rates and dividends rather than about valuation. Treating a widening basis as a valuation signal confuses the carry component with the price component.<\/p>\n<p>Desks that quote a pre-open P\/E off the e-mini are mixing a continuously traded numerator with a quarterly-updated denominator. That is a legitimate thing to do, provided the published number says so, because the resulting series is a price index scaled by a constant between earnings updates. It is not a legitimate thing to compare against a trailing P\/E computed at a prior close on a different denominator vintage.<\/p>\n<h2 id=\"a-reconciliation-checklist-before-you-publish-a-p-e-number\">A reconciliation checklist before you publish a P\/E number<\/h2>\n<p>Before any index P\/E leaves your desk, seven things should be written down next to it. None of them is optional, and together they are the difference between a number another team can reproduce and a number they can only argue with.<\/p>\n<ul>\n<li><strong>As-of date and update time.<\/strong> A ratio built from a noon holdings file and a 4pm index level is not an as-of-close figure.<\/li>\n<li><strong>Aggregation rule.<\/strong> State aggregate, weighted harmonic or weighted arithmetic explicitly.<\/li>\n<li><strong>Earnings basis.<\/strong> GAAP or operating, and whether adjusted items are included.<\/li>\n<li><strong>Horizon.<\/strong> Trailing twelve months, forward twelve months, or a blend, with the blend weights stated.<\/li>\n<li><strong>Loss-maker treatment.<\/strong> Included at negative values, floored at zero, or excluded with weights renormalised.<\/li>\n<li><strong>Identifier-level mapping.<\/strong> Map on an instrument identifier such as a FIGI, and log every unmapped line rather than silently dropping it.<\/li>\n<li><strong>Retained inputs.<\/strong> Keep the holdings snapshot and the divisor alongside the output, and re-run the prior period to confirm the series reproduces.<\/li>\n<\/ul>\n<p>What a holdings-and-fundamentals subscription contributes to this, as distinct from a terminal, is the bottom two layers: the dated holdings file with identifiers, and the archive that lets you re-run a historical date on today&#39;s specification. A terminal gives you a computed ratio quickly and gives you analyst context around it, which is genuinely valuable when the question is interpretive rather than reproducible. <a href=\"https:\/\/americanetp.com\/forum\/2026\/09\/04\/factset-alternatives\/\">FactSet Alternatives: Split by the 4 Layers You Pay For<\/a> separates those layers in detail, and <a href=\"https:\/\/americanetp.com\/forum\/2026\/09\/11\/americanetp-vs-refinitiv\/\">AmericanETP vs Refinitiv (LSEG) 2026<\/a> covers where the two models diverge for an index-reconstruction workflow specifically.<\/p>\n<p>For teams evaluating the analytics layer on its own terms rather than the holdings layer, <a href=\"https:\/\/americanetp.com\/forum\/2026\/09\/14\/refinitiv-review\/\">Refinitiv Review 2026<\/a> and <a href=\"https:\/\/americanetp.com\/forum\/2026\/09\/24\/refinitiv-vs-morningstar\/\">Refinitiv vs Morningstar: 2026<\/a> address what each platform is built for. The honest summary is that a terminal answers questions faster and a holdings subscription answers them reproducibly, and a desk that publishes ratios externally needs the second property more than the first.<\/p>\n<p>AmericanETP updates holdings twice daily, with the primary update at 6pm EST and a secondary update around noon EST, across 3,878 US and global indexes and US-traded ETFs, with FIGIs in the holdings files and archived files back to July 2009. The company was founded in the 1990s as MasterDATA, predating most current ETF data providers, and renamed to AmericanETP to avoid confusion with IBM&#39;s use of the earlier term. Sample files and firm-level enquiries are handled through the AmericanETP site.<\/p>\n<h2 id=\"faq\">FAQ<\/h2>\n<h3 id=\"what-is-the-current-p-e-for-the-s-p-500\">What is the current P\/E for the S&amp;P 500?<\/h3>\n<p>There is no single current figure, because the answer depends on the aggregation rule, the earnings basis, the horizon and the loss-maker treatment. Any number quoted without those four specifications cannot be reconciled against another source. The reproducible approach is to fix an as-of date, pull the holdings file for that date, apply your own stated convention, and keep the inputs alongside the output.<\/p>\n<h3 id=\"what-is-a-good-pe-ratio-for-the-s-p-500\">What is a good PE ratio for the S&amp;P 500?<\/h3>\n<p>&quot;Good&quot; is not a property of the ratio, because the ratio compresses growth expectations, discount rates, sector mix and accounting convention into one number. A reading that looks high relative to a 25-year average may reflect a shift in index composition toward sectors that have always carried higher multiples, with no constituent re-rating at all. The useful question is whether the current reading differs from its history because of within-name multiple changes or because of weight shifts, which a two-snapshot decomposition answers directly.<\/p>\n<h3 id=\"is-the-s-p-500-currently-overvalued\">Is the S&amp;P 500 currently overvalued?<\/h3>\n<p>That depends entirely on the benchmark distribution chosen, and the choice of a 25-year window versus a 50-year window changes the conclusion. Dispersion bands on a forward P\/E chart describe where today&#39;s reading sits in a sample, not whether the index is mispriced. Treating a band crossing as a valuation verdict assumes the sample mean is stationary, which is an assumption about index composition and estimate methodology rather than about markets.<\/p>\n<h3 id=\"what-is-the-highest-p-e-ratio-ever-recorded-for-the-s-p-500\">What is the highest P\/E ratio ever recorded for the S&amp;P 500?<\/h3>\n<p>The record readings on a trailing GAAP basis came from the denominator collapsing rather than the numerator rising, most sharply during the 2008 to 2009 earnings trough, when large write-downs pushed aggregate index earnings close to zero. That is the arithmetic of the aggregate ratio: as the earnings sum approaches zero, the ratio diverges regardless of price. Any historical maximum should be quoted with its basis, because the GAAP series and the operating series peak at very different levels on very different dates.<\/p>\n<h3 id=\"why-does-the-trailing-p-e-differ-from-the-forward-p-e\">Why does the trailing P\/E differ from the forward P\/E?<\/h3>\n<p>The trailing ratio uses reported earnings for the past twelve months and the forward ratio uses consensus estimates for the next twelve, so the two differ by whatever growth is expected plus whatever bias sits in the estimate series. The forward ratio is normally lower, because estimates typically exceed the trailing figure. The forward ratio also falls mechanically when analysts raise estimates faster than prices move, which is a denominator event rather than a repricing.<\/p>\n<h3 id=\"how-do-loss-making-companies-affect-the-index-p-e\">How do loss-making companies affect the index P\/E?<\/h3>\n<p>Under the aggregate rule, a loss-making holding subtracts from the earnings sum, which shrinks the denominator and raises the index P\/E. Excluding loss-makers lowers the ratio and understates it relative to the stated index methodology. Under a weighted arithmetic mean, a holding with earnings near zero produces an individual P\/E large enough to dominate the index figure on its own, which is why that method is not used for index-level reporting.<\/p>\n<hr>\n<p>Rebuilding an index P\/E on your own specification starts with a dated holdings file that carries identifiers and shares outstanding, plus an archive deep enough to re-run the prior period. <a href=\"https:\/\/americanetp.com\/?utm_source=blog-s-p-p-e&#038;utm_medium=blog\">See what AmericanETP&#39;s holdings and fundamentals files contain<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>How the S&#038;P 500 P\/E ratio is actually constructed, why vendor numbers disagree, and how to rebuild it from a holdings file.<\/p>\n","protected":false},"author":1,"featured_media":99,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[4],"tags":[27,108,17,21,14,15],"class_list":["post-100","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-fundamentals","tag-etf-holdings","tag-fundamentals","tag-index-data","tag-market-data","tag-sp-500","tag-valuation"],"_links":{"self":[{"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/posts\/100","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/comments?post=100"}],"version-history":[{"count":0,"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/posts\/100\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/media\/99"}],"wp:attachment":[{"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/media?parent=100"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/categories?post=100"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/americanetp.com\/forum\/wp-json\/wp\/v2\/tags?post=100"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}