Commodity Quant Strategy 04 January 2018 www.globalmarkets.bnpparibas.com 1 MarFATM Metals Monthly; CNY and China continue to lead Over the last 35 weeks, MarFATM has identified CNY on a trade weighted basis, the Baltic dry index, and China government yields as the most important drivers for base metal prices (almost unchanged from the last report).For last 35 business days: China 1y lending rate, speculative positions, and warehouse stocks have been the main underlying factors.The first market component has explained 91.2% of base metal prices over the last 35 weeks (up from 80.1% in Q3 2017).MarFATM is not showing any significant divergence to market prices.We recommend complementing the conclusions of MarFATM Metals with the following links: Commodities and the Fed: should we be worried?Commodities and the Fed: should we be worried? Copper: A tale of two forcesCopper: A tale of two forces Iron ore: Look at the RMB and see the futureIron ore: Look at the RMB and see the future Base Metals MarFATM Weekly PCA methodology is our framework for monitoring market dynamics on a weekly basis, and determining whether commodities are mispriced in the short term. Compared to the publication of October 2017 (first one), the explanatory power of the first component (PC) went up in all the cases (to 91.2% from 80.1% on average). The second and third market factors are becoming nil for base metals prices (Tables 1 and 2, first and second columns). Tables 1-2: Base Metals prices and the three main factors 1st Market2nd Market3rd Market ComponentComponentComponent LME 3m Aluminum86.4%9.3%4.1%1st market factor explains 91% of weekly LME 3m Zinc98.1%0.0%0.0%variations of base metals future prices LME 3m Copper95.0%0.8%3.2%2nd and 3rd market factors LME 3m Nickel87.6%5.4%5.3%have lost explanatory power. LME 3m Lead89.0%10.4%0.3% Average91.2%5.2%2.6%Results as of 12 October 2017 LME 3m Aluminum73.2%18.9%7.3% LME 3m Zinc96.3%0.0%0.8% LME 3m Copper87.4%3.1%3.4% LME 3m Nickel62.5%0.9%28.9% LME 3m Lead81.3%16.5%1.8% Average80.1%7.9%8.5%Source: Bloomberg LLP, BNP Paribas; as of 3 January 2018; PCA estimated on 35 weeks of data Identifying market drivers The PCA process we use to estimate the market factors provides the weekly data series to which metals co-vary, but it does not indicate what is driving these factors. The way we solve this problem is to calculate the R2 between the principal components and a set of 30 weekly market variables; the results are shown in Tables 3 and 4. MarFATM shows that, over the last 35 weeks, the first market component has had the highest co- variance with (a) CNY on a trade weighted basis; (b) the Baltic dry index (proxy of global growth COMMODITY QUANT STRATEGY Please refer to important information at the end of the report This document has been produced by: Banco BNP Paribas Brasil S.A. Gabriel Gersztein Commodity Quant Strategy Head of GM Latam Strategy +55 11 3841 3421 Michael Sneyd Cross Asset Strategy Global Head of FX Strategy +44 20 7595 1307 Gustavo Mendonca Commodity Quant Strategy FX and (c) China 10y government bond yield. Results are unchanged from the last publication. Albeit at low levels, US consumers comfort (also a proxy of US economic performance) and 5y5y US real rates have shown the highest co-variance with the second market component. The explanatory power of the third market component is nil. Tables 3-4: Ranking of top variables with the highest daily R2 to 1st and 2nd principal market components R2 with 1st componentR2 with 2nd component Current6m agoDiffCurrent6m agoDiff Average (5 first variables)65.1%Average (5 first variables)15.7% CNY Trade Weighted Index83.3%3.8%79.6%US Consumer Comfort Index24.9%21.5%3.5% Baltic Dry Index77.4%16.5%60.9%US 5y5y real rate24.5%0.0%24.5% China Government 10y62.7%59.5%3.2%China Data Pulse 11.1%11.6%-0.5% US 1y1y forward53.1%75.2%-22.1%Global Financial Conditions9.4%0.4%9.0% UST Slope 2s5s48.8%54.1%-5.3%China Weekly total Steel Inventory8.7%9.5%-0.8% Warehouse Stocks Cons Index47.2%3.1%44.1%Shanghai Fut Exch Aluminum Deliv Stocks7.3%8.6%-1.3% China Weekly total Steel Inventory46.3%5.8%40.5%China Government 10y6.8%27.4%-20.6% China Economic Surprise 43.0%9.8%33.2%Atlanta Fed GDP 6.6%52.6%-46.0% US dollar broad index41.3%30.3%11.0%Consolidated Spec position5.9%1.8%4.1% BNP Paribas Global Risk40.7%2.5%38.2%St Louis Fed Financial Stress Index4.8%40.7%-35.9% China 5y Swap38.1%63.5%-25.4%US 1y1y forward4.1%1.3%2.8% Shanghai Fut Exch Zinc Deliv Stocks29.1%7.1%22.0%US Economic Uncertainty Policy4.0%0.3%3.7%Source: Bloomberg LLP, BNP Paribas; as of 3 January 2018; PCA estimated on 35 weeks of data Charts 1-9 plot selected components and the explanatory variables with the highest co-variance with them: the R2 between the first component and CNY is back at the highs (83.3%). Charts 1-9: Selected PCs and rolling co-variance (R2) of principal components w/selected explanatory variables -0.45-0.1