Timing the turn – serious signals, or a sentimental journey?

• Despite the view financial flows have a larger impact on equity market valuations than widely acknowledged, there is little compelling evidence in our market data that investment sentiment or financial flows are good predictors of these turning points. 
• Indeed, correlation between annual equity flows and returns for the following year are statistically insignificant. 
• Similarly, investment fund equity exposure is often cited as a good guide to turning points. This reached more than 40% since 2020, and 45.53% at end-2025 — a 30 year high, but that doesn’t mean it can’t go higher. 
• These exposures are also an effect of revaluation of the asset class, and decline in the value of govt bond holdings, rather than any kind of mechanical guide to turning points. 
• Given these issues, we assess Financial Conditions Indicators (FCIs) as a guide to market turning points. Since the GFC, FCIs have become popular with both investors and policy-makers alike as a guide to market performance and financial stability. 
• The FTSE Russell US FCI back-tests well against US equity market performance and both the scale and rate of change of the FCI are important. 
• This does not mean FCIs can be relied upon mechanically to predict equity market turning points, but does suggest they provide valuable information for investors.

Market participants seem increasingly focused on whether equity markets are in bubble territory and, if so, how far. For example, a 950%-plus increase in “AI bubble” Google searches was recorded between late 2024 and late 2025. This paper doesn’t attempt to call that, but instead reviews certain metrics used to identify equity market peaks and troughs.

Firstly, we’ll briefly look at popular measures of investor sentiment. While it’s important to mention that these aren’t intended as tools to identify such turning points, they are often interpreted as such. 

Two indicators that are often mentioned in the media – Investors Intelligence (chart 1, top) and AAII (bottom) –are surveys where investors indicate whether they are bullish or bearish. Strategists say if the ratio between those who are bullish vs. those who are bearish (bull/bear ratio) is very high (see black dashed lines as long-term averages), then this can signal a ‘pull-back’ in the S&P 500.  But when plotted these against the S&P 500, it doesn’t appear to do all that good of a job in calling market turns. There is no identifiable point where “bullish” becomes “too bullish”. 

Investment fund total net assets (TNA) and flows are also referenced as harbingers of equity market turning points. We’ll first take equity exposure is a leading indicator of out- and underperformance for the asset class. 

Using LSEG Lipper data for all European-domiciled mutual funds and ETFs for the century so far, we looked at the relationship of both total net assets and funds flows relative to the performance of both the FTSE All World TR EUR index and Lipper’s Equity Global fund classification. 

Some things do tally. For example, the last equity peak as a percentage of all fund assets was in 2006-7 (41.44%), prior to the global financial crisis. For 2006-11, the five-year return for the FTSE All World TR EUR was -4.65%, and for Equity Global, -12.69%. The average rolling five-year return between 2000 and 2020 was 47.99% for the FTSE All World TR EUR and 30.06% for the Lipper’s Equity Global index. (see chart 2). 

Equity exposure has been more than 40% since 2020, and at the end of 2025 was 45.53%— the highest for the past 30 years (chart 3). But clearly, if you had taken the previous pre-GFC peak as a sell signal, you’d have missed a lot of growth. 

Recent equity exposure is high by historic standards, but that doesn’t mean it can’t go higher. These turning points are an effect of revaluation rather than any kind of useful indicator: as equity markets recover, values increase and the weight of equities within the market portfolio will therefore increase. More investors get on board, as bears morph to bulls, amplifying the effect. On the other hand, as equity valuations crash, their weight within the market portfolio will decrease, encouraging more investors to jettison equities. 

Since neither sentiment, fund assets nor flows are great predictors of equity market turning points, logical alternatives to consider are financial conditions indicators (FCIs). It is helpful to bear in mind how recent interest in FCIs developed, to appreciate why they may be useful in explaining, and even predicting market turning points. Before the GFC of 2008, the prevailing orthodoxy was that financial markets were largely self regulating, allocated resources efficiently, and rarely impacted economic activity. This was grounded in the neo-classical paradigm, and based on the view market forces and the price mechanism are the most efficient way of generating economic growth without the need for central state planning. 

A relatively long period of stable growth and low inflation in the G7 economies – sometimes referred to as the Goldilocks era or “Great Moderation” – from the early-1990s to 2008 reinforced the view that financial conditions were generally stable. Although Japan was mired in a deflationary era of stagnant nominal GDP, it was assumed an exception, caused by a boom/bust in property and dysfunctional banking system. Keynes’ liquidity trap and self-feeding contractions in economic activity during the 1930s were a distant memory. Light touch financial regulation predominated since it was assumed market participants were incentivised to self-regulate to prevent reputational risk.

Since financial conditions cannot be summarised in one variable, like policy rates, putting the concept to work requires a composite indicator. One of the best known FCIs is the Chicago Fed National Financial Conditions Indicator, launched in 2011, and back-tested to 1971. This has 105 variables in total, with three major sub-indexes, for credit (33 variables), risk (33 variables) and leverage (39 variables). It is built as a dynamic factor model so many of the variables are strongly correlated, but the data requirements are substantial and it is harder to identify an underlying investment narrative in FCIs with large numbers of variables, and may be vulnerable to regime shift. The time series is also dominated by the enormous spike during the GFC, as chart 4 shows. 

Our own FTSE Russell FCIs use a more pared-back approach, based on seven subsets of macro variables, which are equally weighted and intuitively connected to financial conditions. Equities are included in the FTSE Russell FCIs, since they are a key component in the cost of capital, but only equity market volatility, and not equity market levels, to avoid circularity. The variables are standardised by Z-scoring and then combined to give an overall composite indicator value, centred at zero. We also avoid so-called “look-ahead bias” by ensuring the Z-scoring at time “t” only captures data released before time “t”. By using a smaller set of variables, we were able to build FCIs for the US, Eurozone, Japan, UK, Canada and China. 

Clarity is required on the objectives of an FCI. Central banks use FCIs to help assess whether financial systems are stable, and as an aid to setting monetary policy, in contrast to investors who may wish to use them for market assessment and forecasting. Our intention was not to produce a forecasting tool for financial markets, since FTSE Russell is an index provider and not a forecasting house. Instead, the objective was to capture financial conditions at a given point in time, and to see how these mapped to financial market performance. Back-testing the FTSE Russell FCIs, we find they generally capture major financial and economic dislocation well, as shown in Chart 5.

Extending this to US equity market turning points, Chart 6 shows some evidence of financial conditions tightening in advance of the sell-offs in the Russell 1000 during the GFC, and Covid, Ukraine and tariff tantrum episodes. But the speed of the movement, as well as the scale of the change in financial conditions appears key, with very rapid tightening a key determinant. 

The Chart also suggests there may be threshold effects in FCIs, whereby FCIs only become critical for equity markets when the scale of the moves exceed certain ranges. In terms of standard deviations, the period before the GFC, Covid and Ukraine shocks shows a tightening of about 1.25 to 1.5 st.deviations from the mean caused significant corrections in the Russell 1000, whilst the tightening before the US tariff shock was less than 1 st.deviation. Standard deviation moves in FCIs of less than 0.75 have not been followed by major equity market turning points. 

It is a similar story when FCIs loosen, as opposed to tightening, in terms of the scale of the moves after shocks, and its speed. The Russell 1000 was slower to recover after the GFC but, as central bank reaction functions became better known, since the GFC, and markets became aware QE asset purchases are likely, recoveries in risk-on assets have also accelerated.