*Abstract* Projects that aim to Reduce Emissions from tropical Deforestation and Degradation (REDD+) have great potential to mitigate climate change and biodiversity loss, but substantial funding is needed to scale up efforts. The trade of carbon credits, quantified as the amount of avoided carbon emissions relative to a baseline ("counterfactual"), can be a key finance mechanism but face numerous challenges. We address the challenges of 1) evaluating the method used to estimate counterfactuals and 2) producing reliable forecasts of carbon outcomes in prospective projects with two studies, using remotely sensed forest loss data and pixel matching to track deforestation and carbon loss trends. In the first study, we evaluated counterfactual-estimating methods with placebo "projects", where there are no REDD+ activities and where we project and counterfactual outcomes are expected to follow the same trend. We found that the ex-post method (estimates made after project start) outperformed the ex-ante methods (forecasts made at project start), supporting the use of ex-post methods for credit issuance and showcases the potential of using the placebo approach to help develop more credible counterfactual-estimating methods. In the second study, we used historical carbon loss to generate forecasts of counterfactual carbon loss after project start, and constructed predictive models for within-project carbon loss and carbon credit production (difference between project and counterfactual carbon loss), using factors theorised to influence REDD+ project effectiveness (slope, remoteness, project size, GDP, corruption index) as predictors. Predictions for both counterfactual carbon loss (goodness-of-fit: 0.62) and within-project carbon loss (goodness-of-fit: 0.87) performed reasonably well, but the predictive performance of carbon credit production were low (goodness-of-fit: 0.32), suggesting a mismatch between the prediction of project vs. counterfactual carbon losses or unknown biases that require further research. *Bio* E-Ping is a third-year postdoc, working with Keshav and Professor David Coomes (Plant Sciences) on using satellite data to quantify the benefit of emissions reduction and its permanence in tropical forest conservation projects in the Reducing Emissions from Deforestation and Forest Degradation (REDD+) framework, with the aim of improving credibility of conservation finance mechanisms through carbon markets.
