StarZero All shows Try the podcast clipper →

May 29, 2014 · 20m · mad

Renaud Laplanche, Lending Club // Data Driven #27 // May 2014 (Hosted by FirstMark Capital)

Renaud Laplanche · 17m spoken Mario Baldi · 37s spoken Matt Turck · 1s spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

In this Data Driven NYC presentation, Lending Club CEO Renaud Laplanche explains how online marketplace lending reduces financial intermediation costs and details how machine learning, behavioral analytics, and natural language processing are deployed for automated fraud detection.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 0.1% of the talking time here. How this is scored →

Matt as informed peer 0.0 Guest teaching 0.4 Guest disagreement 0.0 Matt pushing back 0.0
05100:0010:0020:001:21–5:26 · Matt as informed peer 0/10 Lower Intermediation Costs and Value Creation Renaud presents a monologue detailing Lending Club's business model, lower intermediation costs, and reduced operating expense ratio compared to traditional banks. As a solo presentation segment without host participation, host expertise and pushback scores are zero.5:26–8:34 · Matt as informed peer 0/10 Growth Trajectory and Managing Operational Risk Renaud outlines Lending Club's rapid growth rate and operational risk management while introducing why he chose to share fraud detection data. Host metrics remain at zero due to the presentation format.8:34–13:05 · Matt as informed peer 0/10 Fraud Predictors and Device Data Analysis Renaud breaks down device metrics, operating systems, and online footprint predictors used to identify fraudulent applications. As a monologue segment, host metrics are recorded at zero.13:05–17:05 · Matt as informed peer 0/10 Machine Learning, NLP, and Fraud Reduction Impact Renaud concludes his deck by covering machine learning models, natural language processing on loan applications, and falling fraud attempt rates. The segment is entirely monologue, maintaining zero host scores.17:05–20:52 · Matt as informed peer 0/10 Audience Question and Answer Session Audience members ask technical questions regarding third-party Facebook data, fraud verification metrics, and proxy IP detection. Renaud provides detailed, educational responses while host Matt Turck makes only brief introductory/closing remarks.1:21–5:26 · Guest teaching 0/10 Lower Intermediation Costs and Value Creation Renaud presents a monologue detailing Lending Club's business model, lower intermediation costs, and reduced operating expense ratio compared to traditional banks. As a solo presentation segment without host participation, host expertise and pushback scores are zero.5:26–8:34 · Guest teaching 0/10 Growth Trajectory and Managing Operational Risk Renaud outlines Lending Club's rapid growth rate and operational risk management while introducing why he chose to share fraud detection data. Host metrics remain at zero due to the presentation format.8:34–13:05 · Guest teaching 0/10 Fraud Predictors and Device Data Analysis Renaud breaks down device metrics, operating systems, and online footprint predictors used to identify fraudulent applications. As a monologue segment, host metrics are recorded at zero.13:05–17:05 · Guest teaching 0/10 Machine Learning, NLP, and Fraud Reduction Impact Renaud concludes his deck by covering machine learning models, natural language processing on loan applications, and falling fraud attempt rates. The segment is entirely monologue, maintaining zero host scores.17:05–20:52 · Guest teaching 2/10 Audience Question and Answer Session Audience members ask technical questions regarding third-party Facebook data, fraud verification metrics, and proxy IP detection. Renaud provides detailed, educational responses while host Matt Turck makes only brief introductory/closing remarks.1:21–5:26 · Guest disagreement 0/10 Lower Intermediation Costs and Value Creation Renaud presents a monologue detailing Lending Club's business model, lower intermediation costs, and reduced operating expense ratio compared to traditional banks. As a solo presentation segment without host participation, host expertise and pushback scores are zero.5:26–8:34 · Guest disagreement 0/10 Growth Trajectory and Managing Operational Risk Renaud outlines Lending Club's rapid growth rate and operational risk management while introducing why he chose to share fraud detection data. Host metrics remain at zero due to the presentation format.8:34–13:05 · Guest disagreement 0/10 Fraud Predictors and Device Data Analysis Renaud breaks down device metrics, operating systems, and online footprint predictors used to identify fraudulent applications. As a monologue segment, host metrics are recorded at zero.13:05–17:05 · Guest disagreement 0/10 Machine Learning, NLP, and Fraud Reduction Impact Renaud concludes his deck by covering machine learning models, natural language processing on loan applications, and falling fraud attempt rates. The segment is entirely monologue, maintaining zero host scores.17:05–20:52 · Guest disagreement 0/10 Audience Question and Answer Session Audience members ask technical questions regarding third-party Facebook data, fraud verification metrics, and proxy IP detection. Renaud provides detailed, educational responses while host Matt Turck makes only brief introductory/closing remarks.1:21–5:26 · Matt pushing back 0/10 Lower Intermediation Costs and Value Creation Renaud presents a monologue detailing Lending Club's business model, lower intermediation costs, and reduced operating expense ratio compared to traditional banks. As a solo presentation segment without host participation, host expertise and pushback scores are zero.5:26–8:34 · Matt pushing back 0/10 Growth Trajectory and Managing Operational Risk Renaud outlines Lending Club's rapid growth rate and operational risk management while introducing why he chose to share fraud detection data. Host metrics remain at zero due to the presentation format.8:34–13:05 · Matt pushing back 0/10 Fraud Predictors and Device Data Analysis Renaud breaks down device metrics, operating systems, and online footprint predictors used to identify fraudulent applications. As a monologue segment, host metrics are recorded at zero.13:05–17:05 · Matt pushing back 0/10 Machine Learning, NLP, and Fraud Reduction Impact Renaud concludes his deck by covering machine learning models, natural language processing on loan applications, and falling fraud attempt rates. The segment is entirely monologue, maintaining zero host scores.17:05–20:52 · Matt pushing back 0/10 Audience Question and Answer Session Audience members ask technical questions regarding third-party Facebook data, fraud verification metrics, and proxy IP detection. Renaud provides detailed, educational responses while host Matt Turck makes only brief introductory/closing remarks.

speaking balance: gold is Matt, purple is the guest (3 minute bins)

0:00 · Matt 0% · guest 100%0:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%15:00 · Matt 0.5% · guest 99.5%15:00 · Matt 0.5% · guest 99.5%18:00 · Matt 0.2% · guest 99.8%18:00 · Matt 0.2% · guest 99.8%
Sharpest disagreement ▶ 6:30 Internal refusal from marketing and credit executives

Renaud humorously explains how his chief marketing officer and chief credit officer both flatly refused to allow him to share their data publicly.

 Hardest push from Matt ▶ 17:20 Audience question challenging data collection and fraud labels

Audience member Mario Baldi challenges Renaud on how Lending Club accesses user Facebook data and whether flagged fraud attempts are actually verified.

 Biggest teaching moment ▶ 17:52 Explaining fraud metric estimates and false positive trade-offs

Renaud educates the audience on how reported fraud attempts are conservative internal estimates that include false positives to protect investor capital.

 Matt holds his own ▶ 20:07 Explaining true IP identification through proxy servers

Renaud demonstrates Lending Club's technical sophistication by explaining how they detect true IP addresses even when applicants attempt to hide behind anonymizer proxy servers.
the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
Lower Intermediation Costs and Value Creation 0000 Renaud presents a monologue detailing Lending Club's business model, lower intermediation costs, and reduced operating expense ratio compared to traditional banks. As a solo presentation segment without host participation, host expertise and pushback scores are zero.
Growth Trajectory and Managing Operational Risk 0000 Renaud outlines Lending Club's rapid growth rate and operational risk management while introducing why he chose to share fraud detection data. Host metrics remain at zero due to the presentation format.
Fraud Predictors and Device Data Analysis 0000 Renaud breaks down device metrics, operating systems, and online footprint predictors used to identify fraudulent applications. As a monologue segment, host metrics are recorded at zero.
Machine Learning, NLP, and Fraud Reduction Impact 0000 Renaud concludes his deck by covering machine learning models, natural language processing on loan applications, and falling fraud attempt rates. The segment is entirely monologue, maintaining zero host scores.
Audience Question and Answer Session 0200 Audience members ask technical questions regarding third-party Facebook data, fraud verification metrics, and proxy IP detection. Renaud provides detailed, educational responses while host Matt Turck makes only brief introductory/closing remarks.

Statements from this episode (12)

Disclosure
Laplanche: Lending Club takes no balance sheet risk on loans
“We don't take a balance sheet risk. The, all the capital comes from the from the investors.”
Renaud Laplanche May 29, 2014 ▶ 0:55
Assertion Supported
Laplanche: Lending Club's expense ratio is under 2% vs banks' 5-7%
“That ratio for most banks is between five and seven percent, so it's really all operating costs divided by total loans outstanding, that's between five and seven percent. The same metric at Lending Club is less than two percent, and it's coming down quarter af…”
Renaud Laplanche May 29, 2014 ▶ 1:22
Assertion Partly supported
Laplanche: US credit cards average 17% while savings yields near 0%
“The cost of the average cost of a credit card in the US, so the credit, average credit card interest rate is just shy of 17%. So that's really the value delivered to borrowers. On the flip side if you sort of go and deposit money on high yield savings account …”
Renaud Laplanche May 29, 2014 ▶ 3:40
Assertion Partly supported
Laplanche: Lending Club averages 12.5% borrower rates and 8% investor returns
“So LendingClub, in contrast, as an average interest rate just over 12 and a half percent and on, on the so that's the value to borrowers and as return on average close to eight percent net return after fees and after credit losses to investors.”
Renaud Laplanche May 29, 2014 ▶ 4:29
Disclosure
Laplanche: Lending Club caps annual growth at 150% to manage risk
“So we've been controlling growth to a place where it's in between hundred and 150%. We could be growing faster based on supply and demand available to us but we've chosen to Sort of grow at the current pace as a matter of sort of controlling risk and especiall…”
Renaud Laplanche May 29, 2014 ▶ 5:35
Assertion Not checkable as stated
Laplanche: Lending Club receives about 9,000 loan applications daily
“There's about nine, 9000 applications coming to us during that day.”
Renaud Laplanche May 29, 2014 ▶ 7:31
Assertion Not checkable as stated
Laplanche: Lending Club faces under 10 fraudulent applications daily, totaling $140,000
“It's less than 10 loans out of a population of 9000. That happens every day, and so it's, it is a needle in a haystack, ah, but it is really material because there's sort of 10 or so fraudulent loan applications represent a total of about 140,000 dollars.”
Renaud Laplanche May 29, 2014 ▶ 8:03
Assertion Not checkable as stated
Laplanche: Lending Club sees higher loan fraud rates at night
“We have a lot more fraudulent application at night.”
Renaud Laplanche May 29, 2014 ▶ 8:41
Assertion Not checkable as stated
Laplanche: Lending Club sees higher loan fraud rates on mobile than PC
“We, we've seen a much higher fraud rate on mobile and tablet than on PC and laptop.”
Renaud Laplanche May 29, 2014 ▶ 10:47
Insight
Laplanche: Fraud detection relies on correlation rather than causality
“In terms of fraud, we're really trying to correlate different data points without necessarily looking for a causality. What matters is, is correlation in this case.”
Renaud Laplanche May 29, 2014 ▶ 11:03
Assertion Not checkable as stated
Laplanche: Lending Club detects higher mobile loan fraud on Android than Apple
“We've seen a higher fraud rate in terms of mobile OS with Android than Apple.”
Renaud Laplanche May 29, 2014 ▶ 11:13
Insight
Laplanche: Fraud deterrence requires being harder to target than competitors
“What really matters is not to be smarter than the fraudsters, it's to be smarter than the next guy”
Renaud Laplanche May 29, 2014 ▶ 16:24
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 conversations transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more